PAGESWAPPER VAX SIG Greetings for 1983 by Ross W. Miller, Chairman We just completed a very successful Symposium in Anaheim, California at the Disneyland Hotel. The hotel was a very nice place to have the Symposium, in spite of the wind that took down the big top. Even though the economy has been rather tight this past six months we still were able to maintain a strong interest in DECUS. The Fall DECUS held at Anaheim boasted over a thousand more members in attendance than the Fall DECUS held in Los Angeles in 1981. The one major problem with the exceptional attendance was a good one. We found ourselves short on space because of not having prepared rooms large enought to accomodate those showing interest. So in addition to it being a successful Symposium, we hopefully have learned some things and will be able to work on our space problems for next time. The DECUS elections took place at Anaheim, and I found myself installed as the new Chairman of the VAX Systems SIG. Because of this, the final days of the Symposium proved to be quite eventful for mas as I started the process of picking up the load that past leaders have handled so well. The size of our SIG creates both opportunities and problems, which we hope to address in the near future. It has been suggested that some of the upcoming sessions be videotaped and/or audiotaped. This would be done so that those who are unable to attend the Symposia could participate and benefit from these sessions. DEC has also expressed a real interest in doing this to help save some of the time they spend attending Symposia around the world. We would like to hear from our members on this. 1 PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX SIG Greetings for 1983 Our goals for the next six months are as follows: 1. Quicker turn-around in the System Improvement Request program. 2. Monthly editions of the Pageswapper. 3. Improving the organization and building on SIG strengths through more involvement by members in Working Groups. The purpose of the Working Groups is to facilitate the interchange between those people who have common interests or problems and to refine issues and reach resolutions where possible and then disseminate that to others. Working Groups may either be temporary or permanent, based on the needs of the particular Working Group. It is hoped that through the Working Groups more people can become involved and they can refine the SIR process. 4. Get the SIG Operating Procedures approved and in operation. 5. Hold elections. Most importantly, the entire DECUS membership is made up of volunteers. These are the people who will step forward and help out by providing their own time, resources, and talents to improve the overall community in which we work. We would like to hear from you any suggestions or ideas you might have. Please use the Newsletter as an interesting and informative vehicle by presenting your ideas and suggestions through it. If you do some up with a good solution to a problem or some unique programming please submit it to the DECUS Library. I look forward to a very exciting and interesting future in working with the VAX SIG and I encourage each of you to become personally involved and committed to doing your part in improving the SIG. Thank you! 2 PAGESWAPPER - February 1983 - Volume 4 Number 4 In this issue... In this issue... VAX SIG Greetings for 1983 . . . . . . . . . . . . . 1 Software Shipment Survey . . . . . . . . . . . . . . 3 Editor's Workfile . . . . . . . . . . . . . . . . . 4 The System Improvement Request Program . . . . . . . 5 Stand-alone Backup on your System Disk . . . . . . . 6 INPUT OUTPUT . . . . . . . . . . . . . . . . . . . . 7 Report on Fall 1982 DECUS Symposium . . . . . . . 15 VAX Systems SIG Spring 1983 SIR List . . . . . . . 107 VAX Systems SIG Spring 1983 SIR Ballot . . . . . . 117 INPUT/OUTPUT Submission Form . . . . . . . . . . . 119 System Improvement Request Submission Form . . . . 121 Material for publication in the Pageswapper should be sent to: Larry Kilgallen Box 81, MIT Station Cambridge, MA 02139-0901 Preference is given to material submitted as machine-readable Runoff source. Software Shipment Survey It is not always obvious when new software from DEC is released, and in some cases sites have missed a layered product version through DEC error, only to have it hold up subsequent installation of some other DEC software due to compatibility problems. This section of the Pageswapper is intended to share the information about receipt of SDC shipments between SIG members. This should be of particular assistance to sites which do not have an active LUG. Product Version LUG Meeting Date Status Pascal 2.0 Boston VAX 11-Jan-1983 nobody has it LUG (two got it the next day) Pascal 2.0 New Mexico 27-Jan-1983 all eligible VAX LUG sites attending (3) have it 3 PAGESWAPPER - February 1983 - Volume 4 Number 4 Software Shipment Survey Send in similar information for future software shipments based on surveys at your LUG meetings. 4 PAGESWAPPER - February 1983 - Volume 4 Number 4 Editor's Workfile Editor's Workfile Last month was a lean one, but this month articles for the Pageswapper started rolling in. Including as much as I can, rather than saving some for a rainy day, may seem foolish at first, but TIMELY communication is what we should be about. If next month's goes thin again you can fix it for the future by convincing those in YOUR LUG who have something special to share that they should write it up and send it in. I was rather disappointed in the note I got back from the DECUS office indicating the schedule for printing, mailing and bureaucratizing the January issue. It was going to take about 4 weeks, which seems a bit long to me. We have been told that there might be a way to speed up the printing turnaround (the longest phase), but for now it looks like you will not be reading these words until the March issue has been turned in to the DECUS office. We hope set up a schedule of themes for each month's Pageswapper, but we do not have headway (as mentioned above) to solicit input for a March theme. So how about April? I would like to feature SECURITY related articles, and I hereby solicit your contributions (machine-readable, by March 25) for that issue. I already have somebody in the Boston area who promises to outline some techniques they use for allowing operators to function without unlimited privilege. I am sure Doug Brown will be supplying material from the security working group. That leaves YOU. For those of you who missed INPUT/OUTPUT submissions, they are back this month. Just after the last issue was submitted I got a large Care package from Jim Ebright, including some back INPUT/OUTPUT submissions. (Don't try sending things to me by UPS, it takes longer.) This issue was prepared with otherwise uncredited assistance from Donald Robbins of Sandia Labs, Cliff Hafen of MIT and Art McClinton of MITRE. 5 PAGESWAPPER - February 1983 - Volume 4 Number 4 The System Improvement Request Program The System Improvement Request Program Gary Grebus SIR Coordinator As the new System Improvement Request coordinator, I am very excited about being a part of this important function of the SIG. The SIR program provides a chance of all of us to influence the future directions of VAX and VMS. DEC is interested in user input and the SIR's are a major vehicle for the expression of our concerns and desires. A number of current VMS features were past top ten SIR ballot items. In this and future issues of the Pageswapper, you will find a SIR Submittal form. Make some copies of it. When you think of some improvement or feature that is needed in the VAX, VMS, documentation, or a layered product, write down your idea and submit an SIR. Site the advantages of your suggestion or give an example of how the change would apply. Don't hesitate because you think your idea has already been suggested. We need everyone's input to make the SIR program work. Twice each year, the SIR's received will be collected into a ballot to be published in the Pageswapper. The contents of the ballot will be prepared with the input and assistance of the SIG Working Groups. The ballot provides the opportunity for the SIG membership to assign a priority or rank to the suggested improvements. DEC then provides a formal response to the top ten ranked SIR's. This response is presented at the DECUS Symposium following the ballot. As you may have noticed, this issue of the Pageswapper contains the SIR ballot for response at the Spring symposium. Please fill out and return the ballot as soon as possible. A long lead time is needed to allow the ballots to be tabulated and to allow time for DEC to prepare a response. I hope to continue to produce a ballot immediately following each symposium. By having two ballots per year, each can kept to a reasonable size. The last issue of the Pageswapper contained the list of the top ten SIR's from the Fall ballot. The final analysis of the ballot and the formal responses from DEC are not yet ready. Look for a complete report on the ballot results in the next Pageswapper. 6 PAGESWAPPER - February 1983 - Volume 4 Number 4 Stand-alone Backup on your System Disk Stand-alone Backup on your System Disk (The following augments the information in section 3.3.4 (page 3-22) of the VAX-11 Utilities Reference Manual for VAX/VMS Version 3.0.) (In the following, it is assumed that the system disk is "DRA0"; as far as I know the procedure will work for any type of disk.) (The following is for a 780, those who have other CPU types will have to adapt it as necessary.) 1. Log in as the system manager. 2. Create a new system root directory for the stand-alone BACKUP: $ CREATE/DIRECTORY SYS$SYSDEVICE:[SYSE] 3. Copy the stand-alone BACKUP files from SYS$SYSTEM to the new directory: $ @SYS$UPDATE:STABACKIT DRA0: [SYSE.SYSEXE] 4. Place the console diskette in the console floppy drive and connect the console (if not already connected) and mount the diskette: $ RUN SYS$SYSTEM:SYSGEN SYSGEN> CONNECT CONSOLE SYSGEN> EXIT $ MOUNT/FOREIGN CSA1 5. Copy the default boot command file from the console diskette so that it can be edited: $ MCR FLX /RS=CS1:DEFBOO.CMD/RT/FA 6. Edit the file DEFBOO.CMD placing the hexadecimal quantity E0000000 into register 5: old line: DEPOSIT R5 0 (or D R5 0 ) new line: DEPOSIT R5 E0000000 (or D R5 E0000000 ) 7. Rename the resultant edited file: $ RENAME DEFBOO.CMD BCKBOO.CMD 7 PAGESWAPPER - February 1983 - Volume 4 Number 4 Stand-alone Backup on your System Disk 8. Copy the new file back to the console diskette: $ MCR FLX CS1:/RT=BCKBOO.CMD/RS/FA Then, to perform a stand-alone BACKUP, do the following: $ @SYS$SYSTEM:SHUTDOWN (after system shuts down, type ) >>> BOOT BCK Stand-alone backup should boot in a few seconds, without any messages calling for diskettes to be loaded. FROM: Software Telephone Support Center VIA: Bill Moore and Alan Zirkle Naval Surface Weapons Center Dahlgren, Virginia 22448 (Alan Zirkle = 703-663-7815) VIA: Art McClinton MITRE Corporation 703-827-6352 INPUT OUTPUT A SIG Information Interchange A form for INPUT/OUTPUT submissions is available at the back of the issue. INPUT/OUTPUT 97 Caption: Read/Write magtape with less than 14 byte record Message: Has anyone found a way to read/write magtapes using record lengths less than 14 bytes long? Preferably with QIO's from Macro. Would like ability to read single byte if possible. Contact: Robert Coenen Code 3114 NAVWEPCEN China Lake, CA 93555 8 PAGESWAPPER - February 1983 - Volume 4 Number 4 INPUT OUTPUT Date: September 8, 1982 INPUT/OUTPUT 98 Caption: Personal Literature Retrieval System Message: We are looking for software to efficiently support a Personal Literature Retrieval System for an R&D organization of several hundred research scientists. Contact: Dr. Joseph D. Andose Merck & Co., Inc. P.O. Box 2000 Rahway, NJ 07065 (201) 574-4391 Date: September 30, 1982 INPUT/OUTPUT 99 Caption: Disk File Archival System Message: We are looking for a File Archival and Retrieval System which will facilitate storage and retrieval of infrequently used disk files on tape. Contact: Dr. Joseph D. Andose Merck & Co., Inc. P.O. Box 2000 Rahway, NJ 07065 (201) 574-4391 Date: September 30, 1982 INPUT/OUTPUT 100 (response to un-numbered IO) Caption: Accessing Unibus Registers on a VAX Message: Thanks to all who replied to my previous I/O on this. Method 1 1) convert UB register to hex 2) add base (e.g. FC0000 for 750) 9 PAGESWAPPER - February 1983 - Volume 4 Number 4 INPUT OUTPUT 3) examine or deposit using physical and word (e.g. E/W/P FFEC16) Method 2 1) >>>SET REL:20100000 (base for 780) 2) >>>SET DEF OCT 3) >>>E/W 772520 (unibus address) 4) >>>D 772520 100 (octal) Note - base addresses are given in appendix G in the "Guide to Writing a Device Driver" and are incorrect in the version 2.5 document. Contact: Paul Wilfong 10975 Torreyana Road, Ste. 300 San Diego, CA 92121 (201) 574-4391 Date: October 11, 1982 INPUT/OUTPUT 101 Caption: ** Wanted ** Copy of Atlanta Symposium VAX/VMS Handouts Message: I'll pay Xerox and shipping charges for a copy of the Atlanta Symposium VAX/VMS Handout. (I received the July 1982 Decuscope too late to order; DECUS says they're all gone. Contact: K. Stephen Tinius HQ USEUCOM Box 741 APO NY 09128 Germany 0711-7301-8439 Date: October 26, 1982 INPUT/OUTPUT 102 Caption: VAX-11 PSI -- Software written during a full moon? 10 PAGESWAPPER - February 1983 - Volume 4 Number 4 INPUT OUTPUT Message: We have recently purchased VAX PSI, the X.25 interface software, and are having some minor problems; it tends to crash the system for example. Does anyone out there have experience/advice/information on using VAX PSI. How about an RTPAD for PSI? Contact: Joel M. Snyder 5055 East Broadway Suite A-110 c/o Compuserve Incorporated Tucson, Arizona 85711 (602) 790-5061 Date: unknown INPUT/OUTPUT 103 Caption: Linear Programming Message: I am looking for a comprehensive and easy to use LP package for the VAX. We already have MINOS but it lacks some of the niceties of the major packages for other machines, e.g. output of price ranges, parametric programming, integer programming. Also matrix generation and report writing would be useful. Contact: A. J. D. Griffiths Computer Centre Lincoln College Canterbury, New Zealand 252-811 Date: November 5, 1982 INPUT/OUTPUT 104 Caption: In response to two I/O's on Unibus interface to VAX Message: In response to two notes in the INPUT/OUTPUT section (one from Robert Rosen on interfacing a VAX to a 68000 using a DR11W, and another from Paul Wilfong on accessing the Unibus from the console), the following experience may be of interest. I have been involved in a project which requires a high-speed interface from a 6800-based camera to a VAX-11/780. We are using a DR11-W configured in link mode and the XADRIVER as distributed. The only communication from the VAX to the 6800 are the READY and BUSY bits, since the data always goes from the 6800 to the VAX in fixed length buffers. We use the C0, C1, WC INC ENABLE and BA INC ENABLE bits to transfer the data in 8-bit bytes 11 PAGESWAPPER - February 1983 - Volume 4 Number 4 INPUT OUTPUT into successive memory locations in the VAX. Unfortunately, the User's Guide to the DR11-W neglects to mention that the odd bytes must be placed in the high order bits 15:8. We happened onto this trick in the DRV-11B description (having observed that the two devices can communicate with one another) shortly before DIGITAL TELEPHONE SUPPORT advised us that the current 1981-1982 Peripherals Handbood had a similar note; the only peripherals handbook which I had access to was the 1980 issue which does not mention it either, as far as I can find. One problem which was rather plaguing early on was the fact that the DR11-W showed up off-line after having been on-line. My RSX experience suggested that the device had been removed and hence did not respond when addressed. However, an EXAMINE/WORD showed it there (use EXAMINE/WORD to avoid the VAX trying a quad-word fetch which the Unibus doesn't like, and use 2013E000 device address offset from 760000 octal - see page 335 of the 1980-1981 VAX Hardware Handbook). It appears that the device driver does some check-out of the device when it is loaded, and since Field Service had left the maintenance wrap-around cable installed, the device showed sick to the driver, and was marked as off-line. Contact: Dan R. Mackie Daniel Computing Systems Incorporated 351 Maitland Hill NE Calgary, Alberta T2A 5V4 (403) 273-2657 Date: October 28, 1982 INPUT/OUTPUT 105 Caption: HASP for VAX available from GEJAC Message: HASP is a generalized communications package installed as a system utility on VAX-11/730, VAX-11/750, VAX-11/780 and VAX-11/782. It can operate as an emulator of any standard HASP multileaving workstation (RJE mode) and it can operat as a host to other workstations (Host mode). The protocol uses bi-directional communications and automatic character compression to provide line efficiencies that are not available with other packages such as the 2780 or 3780. Full support is provided for an operator console plus the maximum concurrent mix from each of seven input, print, and punch streams. Requires a DU or DUP-11, optional hardware supported includes KG-11 and DN-11. DMF-32 support under development. 12 PAGESWAPPER - February 1983 - Volume 4 Number 4 INPUT OUTPUT Contact: Gregg E. Church GEJAC, Inc. P.O. Box 188 Riverdale, MD 20737, USA (301) 864-3700 Date: November 8, 1982 INPUT/OUTPUT 106 Caption: Don't Ask DEC for Trivia Message: The answer to several SIRs is a simple user written program installed with appropriate privilege and check: S82-2 use a com file to write a response to a separate file, S82-5 simple Fortran program calling SYS$BRDCST, SR82-9 a simple Fortran program check requested priority against SYSUAF, SR82-16 simple Fortran program which checks the UICs, SR82-23 hibernate 5 seconds retry, SR82-24 see 82-5, S82-27 again simple Fortran program checks UIC, S82-34 try SET PROCESS/PRIV=NOxxx or call SYS$SETPRV, SR82-76 small program to write acocunting field of process header, etc. Contact: David E. Sterritt, Mail Stop 77 Borg-Warner Corp. Byron Jackson Pump Division Post Office Box 2017, Terminal Annex Los Angeles, CA 90015 Date: September 30, 1982 INPUT/OUTPUT 107 Caption: DN-11 Device Handler for VMS V 3.0 Message: We need a DN-11 deriver for VMS V 3.0 running on VAX-11/750 and 730. If you have on to give or sell please get in touch with me. Contact: Alex Yeung Bell Canada 2 Bloor Street West 32/F Toronto, Canada M5G 1W9 Date: November 11, 1982 13 PAGESWAPPER - February 1983 - Volume 4 Number 4 INPUT OUTPUT INPUT/OUTPUT 108 Caption: Program to read IBM floppy diskette on VAX-11/780 RX02 (like FLINT under RSTS/E) Message: We need to read IBM compatible single-sided, singel density EBCDIC floppies from our RX02's on VAX. Does anyone know of a program that would translate them? If so, please give me a call. Any information would be appreciated. Contact: Rhonda Topper Babcock & Wilcox Research Center 1562 Beeson Street Alliance, Ohio 44601 (216) 821-9110 x420 Date: November 3, 1982 INPUT/OUTPUT 109 Caption: A VAX Driver for an RX01 Floppy using an RX11 Controller Message: Currently VAX has two driver for Floppy units, DXDRIVER.EXE for the Console RX01 on the Q-bus and DYDRIVER.EXE for RX02's on the Unibus. We need a driver to control an RX01 with an RX11 controller card on the Unibus. Contact: Frank Beno Hamilton Standard Windsor Locks, CT 06086 (203) 623-1621 ex-2922 Date: November 3, 1982 INPUT/OUTPUT 110 Caption: SPSS (c) interface to Datatrieve-32 Message: The SPSS (c) Data Analysis System has added an interface to Datatrieve-32, completing release 9.1 of SPSS. The SPSS batch system offers over 40 major procedure from simple tables through multivariate statistics, as well as the Box-Jenkins time series for accurate forecasting. 14 PAGESWAPPER - February 1983 - Volume 4 Number 4 INPUT OUTPUT Contact: Joan Fee 444 North Michigan Chicago, Il 60611 (312) 329-2400 Date: Unknown INPUT/OUTPUT 111 Caption: Microprogramming the VAX-11/750 Message: DEC sells the VAX-11/750 with a User Control Store (UCS) option. This is the KU750-YG system option. DEC does not, however, provide software tools that allow user microprogramming. We are interested in microcoding some scientific applications software and would like to hear from anyone who has any knowledge about microprogramming the 11/750. Contact: Norm Clerman or Ellis Betensky Opcon Associate, Incorporated 250 West 57th Street New York, NY 10019 (212) 245-7524 Date: December 28, 1982 INPUT/OUTPUT 112 Caption: VAX Disk Usage Accounting Message: I work on a VAX-11/780 and need to implement a system to produce accounting figures on disk-usage per user. Contact: T. F. Stevens, VAX Systems Programmer Philips ISA - UK Division 17 Beddington Farm Road Croydon, CR9 4EN, England Date: November 26, 1982 15 PAGESWAPPER - February 1983 - Volume 4 Number 4 Report on Fall 1982 DECUS Symposium Report on Fall 1982 DECUS Symposium Report on Fall 1982 DECUS Symposium December 6 to December 10, 1982 Jointly submitted by Marty Adkins, Michael Carullo, Almon Sorrell Westinghouse Electric Corporation Baltimore, Maryland Edited by Almon Sorrell 18-January-1983 16 PAGESWAPPER - February 1983 - Volume 4 Number 4 Page 17 CHAPTER 1 INTRODUCTION 1.1 Purpose . . . . . . . . . . . . . . . . . . . . . 1-1 1.2 Scope . . . . . . . . . . . . . . . . . . . . . . 1-1 1.3 Organization . . . . . . . . . . . . . . . . . . . 1-2 1.4 Other Material . . . . . . . . . . . . . . . . . . 1-2 1.5 Future Symposia . . . . . . . . . . . . . . . . . 1-3 1.6 DISCLAIMER . . . . . . . . . . . . . . . . . . . . 1-3 CHAPTER 2 VMS PERFORMANCE 2.1 Performance Measurement Techniques for VAX/VMS . . 2-1 2.2 Performance Measurement Tools . . . . . . . . . . 2-2 2.3 VMS Performance Management . . . . . . . . . . . . 2-3 2.3.1 VMS File Design . . . . . . . . . . . . . . . . 2-4 2.3.2 Disk Fragmentation . . . . . . . . . . . . . . . 2-4 2.3.3 Algorithms & Data Structures . . . . . . . . . . 2-5 2.3.4 System Management . . . . . . . . . . . . . . . 2-6 2.3.5 SYSGEN Parameters . . . . . . . . . . . . . . . 2-6 2.4 VAX/VMS Performance Management Seminar . . . . . . 2-7 2.5 RMS Tuning . . . . . . . . . . . . . . . . . . . . 2-8 2.6 VMS Real-time Performance . . . . . . . . . . . . 2-9 2.7 VMS/CPS - Computer Performance System . . . . . 2-10 2.8 VAX 11/782 QUALIFY . . . . . . . . . . . . . . . 2-11 CHAPTER 3 VMS PROBLEMS & WORKAROUNDS CHAPTER 4 VMS MISCELLANY 4.1 VMS Software Development Group . . . . . . . . . . 4-1 4.2 VAX/VMS System Update . . . . . . . . . . . . . . 4-2 4.3 Replacing Object Libraries with Shareable Images . 4-3 4.4 VAXMATH - A Mathematical Software Library for the VAX . . . . . . . . . . . . . . . . . . . . . . . 4-7 4.5 Miscellaneous . . . . . . . . . . . . . . . . . . 4-8 CHAPTER 5 LAYERED PRODUCT LANGUAGES 5.1 APL . . . . . . . . . . . . . . . . . . . . . . . 5-1 5.2 FORTRAN . . . . . . . . . . . . . . . . . . . . . 5-2 5.3 PASCAL . . . . . . . . . . . . . . . . . . . . . . 5-3 5.4 DATATRIEVE . . . . . . . . . . . . . . . . . . . . 5-3 5.4.1 DATATRIEVE Hints and Kinks . . . . . . . . . . . 5-3 5.4.2 DATATRIEVE Future Directions . . . . . . . . . . 5-6 5.5 FMS . . . . . . . . . . . . . . . . . . . . . . . 5-9 CHAPTER 6 TOOLS:VNX 6.1 DEC/MMS - Module Management System . . . . . . . . 6-2 - 15 - PAGESWAPPER - February 1983 - Volume 4 Number 4 Page 18 6.1.1 What DEC/MMS does . . . . . . . . . . . . . . . 6-2 6.1.2 DEC/MMS and DEC/CMS . . . . . . . . . . . . . . 6-3 6.2 DEC/CMS . . . . . . . . . . . . . . . . . . . . . 6-3 6.2.1 Current Status . . . . . . . . . . . . . . . . . 6-4 6.2.2 Future Directions: . . . . . . . . . . . . . . . 6-4 CHAPTER 7 NETWORKING AND COMMUNICATIONS 7.1 DECnet Phase IV . . . . . . . . . . . . . . . . . 7-1 7.2 Ethernet . . . . . . . . . . . . . . . . . . . . . 7-2 7.2.1 Ethernet performance . . . . . . . . . . . . . . 7-2 7.2.2 DEUNA Ethernet UNIBUS controller . . . . . . . . 7-2 7.2.3 DEC Ethernet strategy . . . . . . . . . . . . . 7-3 7.3 DECnet/SNA Gateway . . . . . . . . . . . . . . . . 7-3 7.3.1 Gateway features . . . . . . . . . . . . . . . . 7-4 7.3.2 Gateway components . . . . . . . . . . . . . . . 7-4 7.4 OBSERVER Network Monitor . . . . . . . . . . . . . 7-5 7.4.1 OBSERVER Features . . . . . . . . . . . . . . . 7-5 7.4.2 OBSERVER Limitations . . . . . . . . . . . . . . 7-7 7.5 Miscellaneous network topics . . . . . . . . . . . 7-7 CHAPTER 8 PRO-SERIES PERSONAL COMPUTERS 8.1 Hardware Configuration . . . . . . . . . . . . . . 8-1 8.2 P/OS and Graphics . . . . . . . . . . . . . . . . 8-2 8.3 PRO-300 Series Communications and TMS . . . . . . 8-4 8.4 Software . . . . . . . . . . . . . . . . . . . . . 8-5 8.4.1 P/OS - Professional Operating System . . . . . . 8-5 8.4.2 PRO Applications Development Tool Kit . . . . . 8-6 8.4.3 PRO-BASIC . . . . . . . . . . . . . . . . . . . 8-7 8.4.4 UCSD P-System . . . . . . . . . . . . . . . . . 8-7 8.5 PRO-300 Future Directions . . . . . . . . . . . . 8-8 8.6 COMDEX Product Announcements . . . . . . . . . . . 8-9 8.6.1 Furniture . . . . . . . . . . . . . . . . . . 8-10 CHAPTER 9 HARDWARE 9.1 MIL-Spec Hardware . . . . . . . . . . . . . . . . 9-1 9.2 TEMPEST . . . . . . . . . . . . . . . . . . . . . 9-1 9.3 VAXStation-100 (VS100) . . . . . . . . . . . . . . 9-2 9.3.1 Components . . . . . . . . . . . . . . . . . . . 9-2 9.3.2 Architecture . . . . . . . . . . . . . . . . . . 9-2 9.3.3 Features & Availability . . . . . . . . . . . . 9-3 9.4 LN01 Laser Printer . . . . . . . . . . . . . . . . 9-4 9.5 LP27 High Speed Printer . . . . . . . . . . . . . 9-5 9.6 LA100 Letterwriter/Letterprinter . . . . . . . . . 9-5 9.7 New Mass Storage Architecture/VMS Support . . . . 9-5 9.7.1 Devices . . . . . . . . . . . . . . . . . . . . 9-6 9.7.2 Software Support . . . . . . . . . . . . . . . 9-10 9.8 Terminals . . . . . . . . . . . . . . . . . . . 9-11 9.9 UNIBUS Timing Considerations . . . . . . . . . . 9-11 - 16 - PAGESWAPPER - February 1983 - Volume 4 Number 4 Page 19 9.10 Power Problems . . . . . . . . . . . . . . . . . 9-13 CHAPTER 10 RSX/VMS RELATED TOPICS 10.1 AME - Applications Migration Executive . . . . . 10-1 10.1.1 Current AME Status Relative to RSX . . . . . . 10-1 10.1.2 Future plans for AME . . . . . . . . . . . . . 10-2 10.2 RSX -> VAX: Programming for Transportability . . 10-2 10.3 Mathematics of RSX11M . . . . . . . . . . . . . 10-3 CHAPTER 11 SITE MANAGEMENT AND TRAINING 11.1 SIG Panel . . . . . . . . . . . . . . . . . . . 11-1 11.2 DEC Education & Training . . . . . . . . . . . . 11-2 CHAPTER 12 COMPUTER SECURITY 12.1 Security Issues in VMS Access Control . . . . . 12-1 12.2 Network Security - Proxy Logins . . . . . . . . 12-2 12.3 VMS Security . . . . . . . . . . . . . . . . . . 12-5 12.3.1 Common VMS attacks and counters . . . . . . . 12-5 12.3.2 General tips . . . . . . . . . . . . . . . . . 12-6 12.3.3 Lost data issues . . . . . . . . . . . . . . . 12-7 12.4 Miscellaneous Tidbits on VMS Security . . . . . 12-7 APPENDIX A VAX/VMS SIG FALL 1982 HANDOUTS TABLE OF CONTENTS APPENDIX B NETWORKS SIG FALL 1982 HANDOUTS TABLE OF CONTENTS APPENDIX C DATA MANAGEMENT SIG FALL 1982 HANDOUTS TABLE OF CONTENTS - 17 - CHAPTER 1 INTRODUCTION Purpose This technical report covers the Fall 1982 DECUS (Digital Equipment Computer Users Society) Symposium held in Anaheim, California from December 6 through December 10,1982. It also includes a report on a pre-symposium seminar held December 5, 1982. The report was prepared from input supplied by Marty Adkins, Michael Carullo, and Almon Sorrell, and was edited by Almon Sorrell. The report is not organized chronologically, but rather by topic, since this seemed to be the most useful means of conveying information to the reader. Scope This report addresses the sessions that are of greatest interest to VAX users. It may be interesting to note that nearly 4300 people were in attendance and that there were literally hundreds of formal sessions which were scheduled in 16 different rooms concurrently. Those sessions included presentations on RSX (PDP-11 operating system), RSTS (PDP-11 time sharing), and TOPS (DEC-10/20 operating system) as well as sessions oriented toward the business user, site preparation, etc. Additional informal sessions were held continually in "camp grounds" where users and VMS developers could meet for one-on-one discussions. - 18 - INTRODUCTION Page 1-2 Organization Organization The top level organization of the report consists of 11 major categories: VMS Performance, VMS Problems, VMS Miscellaneous Topics, Languages (APL, Fortran, Basic, Pascal, Datatrieve, FMS,etc.), VNX Tools (MMS, CMS), Hardware, Networks and Communications, Personal Computers & Software, RSX/VMS Related Topics, Site Management & Training, and Computer Security. Other Material There were two volumes of handouts for the VAX/VMS sessions which contained hard copies of the presenters' slides or viewgraphs. Where reference to these are appropriate the symbol "VSIG" will refer to the VAX Systems SIG Handout. Similar handouts were prepared by the RSX SIG (RSIG), Network SIG (NSIG), and Data Management SIG (DSIG). Copies may be obtained through the DECUS office by mailing requests to DECUS International Headquarters, MR2-1/C11, One Iron Way, Marlboro, MA. 01752. All requests must be received by February 25, 1983. The prices below include shipping and handling: o VAX/VMS SIG $ 23.00 o Networks SIG N/A o Data Management SIG 13.00 o RSX SIG 17.00 o Structured Software 13.00 o RSTS 13.00 - 19 - INTRODUCTION Page 1-3 Future Symposia Future Symposia The Spring '83 DECUS symposium will be held at the Cervantes Convention Center in St. Louis, Missouri from May 23 through 27, 1983. The Fall symposium will be held at the MGM Grand Hotel in Las Vegas, Nevada from October 24 through 28, 1983. DISCLAIMER All information reported here is believed to be correct and is based upon information related in good faith to the attendees. Every attempt has been made to check this report for accuracy and correctness, and any additional information gathered since the symposium which bears on topics presented there is also presented. Neither the attendees nor Westinghouse accept any other responsibility for the contents of this report. - 20 - CHAPTER 2 VMS PERFORMANCE Two full-day pre-symposium seminars were given at the Fall DECUS on VAX/VMS performance management and tuning. In addition, there were four or five sessions during the symposium that dealt directly with performance management and many others that referenced topics of performance management. The main theme throughout all of the sessions that we attended was that tuning is responsible only for the last 5-10% of system performance improvement. The major thrust of performance management is in system design and configuration, application design, performance measurement, and capacity planning. Performance Measurement Techniques for VAX/VMS A session was given by Mark Bell of the VMS Performance team entitled, "Performance Measurement Techniques for VAX/VMS". Performance measurement is an important aspect of system management. It permits an educated approach to system tuning to provide a performance improvement. More importantly, this measurement is the major input to capacity planning and new system selection and configuration. There are three types of performance measurement techniques: hardware, firmware, and software. The topic of this seminar was software performance tools, specifically, event-driven and sampling-driven tools. Event-driven tools intercept the flow of control at significant events and record a time stamp. The advantage of an event-driven tool is its generality and flexibility. The disadvantages are the potentially high overhead incurred by this type of monitor and the high development and maintenance costs due to the complexity of the software. Sampling-driven tools consist of monitoring code that is periodically entered to take a snapshot of the system. The advantages of these tools are that they have a controlled overhead and they are easier to add onto an existing system. The disadvantage is that with this method some quantities (e.g. disk I/O time) are not measurable so that some quantities are based on statistical - 21 - VMS PERFORMANCE Page 2-2 Performance Measurement Techniques for VAX/VMS inference. The basic performance items to be measured are: o How busy is the resource (utilization)? o How long does it take to service the request (service time)? o How much time is spent waiting for a resource (queue time)? o What are the processing overlap factors? The VMS group has developed a sampling-driven monitor based on a standard interval timer. It redirects the system control block vector to the measurement code. Data is sampled from unit control blocks, channel request blocks, I/O request packets, and interrupt data blocks. The measurement code records CPU and device busy counts, queue length accumulators, total operation counts, and system state counts. With this information the code can calculate utilization of resources (CPU, memory, and I/O), service time, queue wait times and CPU-I/O overlap percentages. This monitor program was validated with a hardware monitor and shown to utilize less than 5% of the CPU with negligible I/O overhead. The code occupies less than 35 pages of resident memory. The monitor is discussed in detail below. Performance Measurement Tools Mark Bell presented a general session entitled "VAX/VMS Performance Measurement Tools" in which he presented a summary of the measurement tools presently available to VMS managers. The first and the major tool for performance measurement is the MONITOR utility which collects and displays system statistics. It was recommended that MONITOR always be running in background with a 5 minute sampling interval to provide correlation of observed performance problems with system activity. MONITOR is also extremely useful for capacity planning. A log of summarized results will provide a time history of system resource utilization that can point to potential problems in the future as the usage approaches full capacity. The ACCOUNTING utility provides for accounting data reduction and analysis. It provides process or image identification and resource utilization statistics. This information will help in capacity planning by pointing to the images most executed by users and detailing the resources those images use. (RUMOR - the next major release of VMS will provide selective image-level accounting, supposedly with the introduction of a - 22 - VMS PERFORMANCE Page 2-3 Performance Measurement Tools new privilege bit.) The System Dump Analyzer(SDA) can also be used as a performance management tool but should ONLY be used as a last resort. Basically, SDA takes a snapshot of physical memory. It can be used on a running system (ANALYZE/SYSTEM) or a crash dump (ANALYZE/CRASH). SDA produces displays of formatted system data structures such as the PCB, PHD, and JIB. SDA can also be used to investigate MWAITs. That is, to find out which resource a process is blocked on, or to look at I/O request queues. Further performance measurement tools are the DCL commands: SHOW PROCESS/CONTINUOUS, SHOW DEVICE, SHOW MEMORY and SHOW SYSTEM. These commands provide a quick and easy way of seeing the usage of I/O and memory resources and to see what resources are being used by a process. VMS Performance Management A session on "VMS Performance Management" was given by Steve Forgey, another member of the VMS performance management team. There are four major areas of performance management: 1. System Design - includes reducing expected computer usage to individual tasks, estimating costs, and defining a strategy for expansion. 2. System Configuration - selection of hardware and software to match tasks of the overall system and plan for expansion. 3. Appplication Design - optimization of the implementation of the workload. The major areas of effort here are education and training of the application programmers in the subject of VMS and in consultation on major application design efforts. The major components of application design are file design and algorithms and data structures. 4. Performance Aspects of System Management - 23 - VMS PERFORMANCE Page 2-4 VMS Performance Management VMS File Design VMS file design involves trading off space with speed. Space should be considered from the viewpoint of expansion and waste. Speed should be considered from the viewpoint of disk bandwidth and data access. One consideration of note is the contiguity of certain oft-used images. The SYSGEN parameter PFCDEFAULT specifies the number of contiguous blocks to be brought into physical memory when a page fault occurs. This number can only ____ be realized by having the logical blocks of a file be physically contiguous on the disk. Therefore, the full savings of this SYSGEN parameter can only be realized when the file is specified contiguous during the LINK operation or if the file is subsequently copied using COPY/CONTIGUOUS. For images that are not heavily used there is no need to specify a contiguous file during the link, because the savings do not justify the requirement. These images will, however, gain from the PFCDEFAULT setting if the disk cluster size is greater than one. (The cluster size is the number of contiguous blocks that are allocated to a file each time it exhausts the space already allocated.) This will ensure that, for a cluster size of 3, at least three blocks will be brought into memory when a page fault occurs. Disk Fragmentation Another consideration in this area is the fragmentation of the disk. When a file is created, the system will allocate clusters of blocks according to availability as described in the BITMAP.SYS file on the disk. As clusters are returned to the system (through a delete for example) these clusters become available for reallocation. With an empty disk, the allocation of clusters proceeds in a contiguous fashion for all files. Then as clusters are returned, these are allocated first until they are used and then allocation reverts back to the end of the used disk space in a contiguous fashion again. Normally a file will consist of a few groups (extents), each group consisting of a large number of physically contiguous blocks. However, when the majority of files on the disk begin to have many groups, each with only a small number of physically contiguous blocks, the disk is said to be fragmented. This will place an unnecessarily heavy load on the system by requiring continuous head seeks to read a file. This is especially true for image activation. This condition can usually be recognized by noticing the activity of the disk ACP. Usually this ACP consumes an inordinate amount of CPU time when the disk is fragmented. The cure for this condition is to perform regular IMAGE mode BACKUP on the disk and create a new - 24 - VMS PERFORMANCE Page 2-5 VMS Performance Management disk from that backup. Algorithms & Data Structures Another area of application design is algorithms and data structures. Designers should choose algorithms and data structures that minimize the use of "expensive" VMS operations such as process creation, image activation (i.e. applications written in DCL) and short I/O bursts. Some other programming strategies recommended to improve performance were: o Cluster related sections of code o Remove overlays from applications that migrated from other systems o Remove error routines from main code so they do not break up the locality of the oft-referenced code o Reference data in the order it is stored o Initialize data areas just prior to use o Group high-use buffers and data buffers into common areas o When considering block I/O vs. a mapped section approach, remember that caching benefits from random access. For sequential access to a file, block I/O is probably better. Most of these suggestions try to avoid the occurrence of "hard" page faults, which require disk I/O, and try to raise the probability of "soft" page faults, which merely require modification to pointers. System Management In the area of System Management it was again stressed that tuning results in only the last 5-10% of improved system performance. The trend in VMS development is to allow the system to automatically decide the value of the various SYSGEN parameters. Initial configuration should always be done by AUTOGEN. Those parameters that may be adjusted for a unique configuration to avoid failures include: MAXPROCESSCNT, JOBLIM, BALSETCNT,and VIRTUALPGCNT. To avoid serious degradation look at: NPAGEDYN, IRP, SRP, LRP, PAGEDYN, MAXBUF (BYTLM in UAF), SYSMWCNT, LOCKIDTBL and REHASHTBL. SYSMWCNT, - 25 - VMS PERFORMANCE Page 2-6 VMS Performance Management the minimum system working set count, should be adjusted to ensure less than five system page faults per second (less than one fault/second is achievable). Note that by locking the system in memory (disabling POOLPAGING and SYSPAGING), or by preventing system page faults (with a large SYSMWCNT), you may cause a worse impact in process paging. SYSGEN Parameters There are numerous parameters that affect VMS memory management. Memory is the resource that can be managed most by system software. On memory-rich systems, memory can be used to offload I/O and processor usage. On memory-tight systems, memory consumption must be limited to avoid thrashing. Some important parameters for memory management include: PFRATL, the process page fault rate below which pages are taken from a process, and SWAPOUTPGCNT, the number of pages to which each process will be trimmed if pages are still needed by the system after all processes are trimmed to their WSQUOTA. In version 3.1 of VMS, PFRATL is set to zero as the default in order to disable working set trimming. In this case (PFRATL=0) trimming will only occur when the system needs memory (when the number of free pages falls below FREELIM). The rules for setting SWAPOUTPGCNT are different depending on the type of processor. The lower the setting of SWAPOUTPGCNT the more trimming and the less swapping that will occur. The major disadvantage to trimming is that it will increase the rate of page faulting because each process has a smaller working set. SWAPOUTPGCNT must be set to encourage trimming when the cost per page to trim and then fault the page back in, is less than or equal to the cost per page to swap out a process and bring it back in. When this is not true SWAPOUTPGCNT should be set to encourage swapping. The slower the CPU the more trimming should be encouraged. Therefore, on a VAX 11/780 swapping is preferrable to automatic trimming and on a VAX 11/730 the opposite is true. - 26 - VMS PERFORMANCE Page 2-7 VAX/VMS Performance Management Seminar VAX/VMS Performance Management Seminar A pre-symposium seminar entitled "VAX/VMS Performance Management" was given by Mark Bell of the VMS performance group. This seminar was taken from a new course just developed to be offered by DEC's Educational and Training group. The title of the course is "System Performance Management and Capacity Planning" and will be offered at various DEC Education centers during 1983. Some of the topics discussed during the seminar were also touched upon during regular symposium sessions, but in lesser detail. Portions of the student workbook from the above mentioned course were also distributed. The tuning chapters of the VMS documentation set will be rewritten to include a cookbook approach to tuning. This will include a tuning tree to assist managers in troubleshooting system performance problems. An interesting change in the FIFO algorithm for page replacement in the working set was detailed. Initially, when a process's working set is targeted for releasing pages, a FIFO approach was used. With the advent of VMS 3.1 a new SYSGEN parameter TBSKIPWSL comes into play. The new philosphy is that if the page to be returned to the system is one that has just been translated by the system (it is in the translation buffer) it is likely to be referenced again and should be bypassed as the page to be returned. TBSKIPWSL is the number of pages in the working set to bypass before the next page must be returned, regardless of its existence in the translation buffer. This allows circumventing the FIFO scheme at least for the first TBSKIPWSL pages. NOTE According to comments made during the session, hardware ECO level 7 must be installed in VAX 11/780 CPUs to enable use of this new implementation. Otherwise, the system reverts to the standard FIFO approach. An extensive explanation of memory management was given and considerable attention was paid to Automatic Working Set Adjustment (AWSA). As explained above, PFRATL=0 is the new default for VMS 3.1. This is due to the fact that lowering the working set is expensive in overhead, unlike raising the working set. The AWSA's major goal is to substitute "soft" page faults for "hard" page faults. The estimated time for a "soft" fault is 100-200 microseconds while a "hard" fault will take 30 milliseconds. The percentage of hard page faults should be below 15%. Another parameter explained above is - 27 - VMS PERFORMANCE Page 2-8 VAX/VMS Performance Management Seminar SWAPOUTPGCNT. Setting this parameter too low may result in a high number of page faults and no swapping. This will be undesirable for a VAX 11/780. Mention was made of a program available from DEC entitled "DECALC" that is used to determine what effect various job mixes would have on the system. Lack of time prevented the acquisition of any further information. RMS Tuning Peter Lieberwirth, VAX RMS developer, gave a Q&A session on RMS tuning (VSIG pp. 353-358). o CONVERT/FAST bypasses RMS -- use it! The switch exists because CONVERT calls SORT32 which uses beaucoup memory and scratch files. In a few rare cases, one might need to use the /NOFAST. o ANALYZE/RMS currently lacks some of the capability which was in the unsupported VMS 2.5 version. They are looking at restoring those features. o I reported that ANALYZE/RMS did not display the "no backup" file attribute. Peter noted it. o I reported that ANALYZE/RMS/FDL will compute a data compression factor of -14. This causes the FDL editor to display a huge 32-bit positive number. Peter explained that the ANALYZE algorithm can produce percentages that are <0 or >100. The bug, of which he was not aware, is in FDL's display routine. o RMS still contains a long-known, obscure bug: deadlock can occur for two simultaneous file accesses which select primary and secondary keys. The cause is still not known. To prevent deadlock, RMS always traverses the index down and then across. o No RMS ISAM fixes yet exist beyond VMS 3.1. o Guideline for selecting key or data compression in prologue 3 files: it is really intended to shrink large files (2000 blocks) and their index. To perform a lookup, the key of reference is compressed, then the search is performed. If key compression is not selected, a binary search is done. o Setting the WAT bit in the RAB will cause the record stream to wait if a record is locked, instead of aborting with a record lock error. - 28 - VMS PERFORMANCE Page 2-9 VMS Real-time Performance VMS Real-time Performance Steve Forgey of DEC presented a follow-up to his spring session on VMS interrupt latency. New items included: o The maximum latency reduction in VMS 3.0 is due to an improvement in the ERRFMT error logger program. In 2.x it did a 512-byte MOVC5 at elevated IPL. This has been changed to a semiphore mechanism. o Data on the 11/730 was presented - maximum latency is 800 usec! This is caused by the low-cost design of the Integrated Disk Controller (IDC). The IDC does not use DMA for data transfers; it grabs the CPU to do it at microcode level. The graphs showed peaks for the one and two-buffer cases (the IDC has two buffers). The measurement was repeated for an 11/730 with a UDA50 instead of an IDC, giving the expected lower results. o The 11/730 uses an 8085 uP as its interval timer. An MTPR instruction talks to it over a serial bus. o Latency distributions for the cases of heavy disk I/O, single-user terminal I/O, and many-user terminal I/O, all give nearly the same result. A related session on real-time response entitled "VAX 11/780 Real-time Aircraft Simulation" was given by Paul Jernigan of Douglas Aircraft (VSIG pp. 119-139). He discussed the problem of preemption of real-time processes. The main culprit here is DECnet (10's of milliseconds) although terminal activity can also cause impact. Use of the Connect-to-Interrupt feature of VMS does not solve this problem, because it uses the AST mechanism. Their solution was a special I/O driver synchronized to an interrupt from an external clock (e.g. KW11-P). This reduced response jitter to under 1 MS. - 29 - VMS PERFORMANCE Page 2-10 VMS/CPS - Computer Performance System VMS/CPS - Computer Performance System VMS/CPS is a product and a service offered by DEC's Software Services group to analyze VAX performance and help locate bottlenecks which degrade throughput. It is available in two forms: 1. Service - The product is bundled with 32 hours of consulting by a level-3 software engineer. The software is then left on the CPU where the analysis was performed. NOTE This software is a product of Software Services and was not developed nor is it supported by Central Engineering Software. The reason it was initially only offered as a service is that it requires considerable expert interpretation. 2. Software only - The software tool kit is available on a per-CPU license basis, with a license to copy available. The tool kit consists of the following separate components: o SAMPLER - a system-wide analysis tool (MONITOR plus additional capabilities) o DVA - Disk/Volume Analyzer - comprehensive disk analysis program to determine fragmentation and utilization of a disk. o PCSAMPLER - tracks the system PC and shows (on a per/process basis) how much time was spent in which system modules - useful to determine what code is taking kernel time. o Program Analyzer - performs histogram of individual program's PC to determine where it's spending time. This is an outgrowth of the PME program by DEC's Bert Beander which was on an earlier DECUS tape. o TRACE - allows trace of system code at user-defined breakpoints within the system. o SWDRIVER - Software driver which serves SAMPLER & PCSAMPLER. - 30 - VMS PERFORMANCE Page 2-11 VAX 11/782 QUALIFY VAX 11/782 QUALIFY With the introduction of the VAX 11/782 dual-CPU configuration, a site with "the right job mix" can achieve up to 80% increase in throughput for a 40% increase in cost. The problem has been in determining if you have the right job mix. Digital Software Services has been using a new tool called QUALIFY to do just that. In a conversation with its anonymous developer, I learned a few details. The 782 contains a primary processor (PP) which behaves as a normal 780, and a secondary processor (SP) which behaves as a slave processor. The two processors use ONLY shared memory (MS780) in which a single copy of VMS resides. The SP cannot perform any I/O, system services, or any other function that is executed in kernel mode. When it needs to do any operation that requires synchronized access, it must save process context and interrupt the PP. The kernel mode code is then executed by the primary. This shift between processors is called a Secondary processor Transition Event (STE), and is a measure of the extra overhead required for the 782. The VMS scheduler _____ ________ must follow a modified algorithm. It will first attempt to dispatch to the SP, but it cannot simply dequeue the highest priority compute-bound process. It must perform additional tests to ensure that the process has no kernel mode ASTs outstanding, or anything else that would immediately kick it back to the PP. If the process running in the SP causes no STEs, it can not be preempted until quantum expiration. For this reason, it may pay to set QUANTUM larger on a 782. Also, most sites should set their batch queue priority the same as interactive. Because of the STE penalty, it is important to keep at least one compute-bound, low-STE rate process running in the SP. If the job mix consisted of only two users running EDT, considerable thrashing would result. This is why the actual performance improvement factor ranges from 0.85 to ~1.8. The QUALIFY program provides an objective assessment by analyzing the actual job mix on the candidate machine. Data is collected on a per-process basis, and a display is produced which places each process into an "acceptable", "uncertain", or "unacceptable" category. The thresholds are chosen to match the cost/benefit break-even point -- the uncertain case roughly corresponds to 40% improvement. QUALIFY executes the following steps: 1. The main program changes into kernel mode and allocates a calculated amount of non-paged dynamic memory (NPAGEDYN). 2. The collection code and an array area based on MAXPROCESSCNT is copied to this region via MOVC5. - 31 - VMS PERFORMANCE Page 2-12 VAX 11/782 QUALIFY 3. The scheduler is patched so that context switch events may be recorded. The STEs occur in process context so collection on a per-process basis is easy. 4. Event times are recorded in quadwords to the usec level (by reading interval clock register). 5. It uses the total time spent in kernel mode and the number of STEs to determine the average duration of an STE. Note that the classic queueing theory model (arrival rate and service time) applies quite well here. 6. At an operator-specified interval, the main image awakes, reads the array area, and displays the results for that period. - 32 - CHAPTER 3 VMS PROBLEMS & WORKAROUNDS This section covers miscellaneous problems in VMS which were uncovered or elaborated upon during the symposium. Any workarounds or cures are shown where known. o There is a known bug in Standalone Backup which can cause corruption of multi-header files on disk. The situation can occur when performing a BACKUP/RECORD operation using any type of S/A BACKUP (i.e., from floppies, TU58s, disks made via STABACKIT.COM, etc.) In an off-line conversation with DEC developer Marty Jack, he mentioned that the bug was not discovered until it was too late to incorporate the correction in VMS V3.1. Thus, the problem will not be fixed until V3.2. There is supposed to be a patch available from local software support. o The accounting file may show a detached process with a UIC not belonging to any user. The problem may be due to the fact that MTAACP is started with UIC [1,3]. If the ACP goes away, accounting is written against the last user to make use of it. o There is a known problem with the image activator which causes no traceback from detached or subprocesses. o There was a report that the Debugger can no longer deposit a value into a shared global section. The Q&A panel had no response, but noted it. o The terminal driver has a problem when using read with prompt and timeout. The timer is started at the beginning of the write. Thus a long write can cause the service to timeout. o There is currently no way to determine the amount of charge remaining in the (memory) battery backup. o A problem exists with the interlocked queue instructions such that if you remove a packet from an empty queue, you can put junk on the address. - 33 - VMS PROBLEMS & WORKAROUNDS Page 3-2 o Using COPY to write to a device MOUNTed/FOREIGN can kill the home block of the device. o DMF32 topics - There is a problem in the terminal driver that can cause SHUTDOWN to hang for users of DMF32s and DMF32 look-alikes. If a user has a terminal XOFFed (CTRL/S), the second BROADCAST message hangs. The only way to clear the problem is to find the terminal and (CTRL/Q). This is supposed to be fixed in VMS 3.2. - For the DMF32, "SET TERMINAL/PARITY={EVEN|ODD}" will have the reverse effect. This is fixed in VMS 3.2. - Formatted terminal output () is optimized in VMS 3.2. - The maximum DMA buffer size for a terminal output is 512 bytes. The DMF driver allocates two UNIBUS mapping registers per line. - On terminal input, the DMF32 will interrupt when the input silo contains a certain number of characters, or when a programmable delay has expired after the receipt of the first character. The delay is set by SYSGEN parameter TTY_SILOTIME. Leave it alone! - Don't leave terminal lines unconnected. - 34 - CHAPTER 4 VMS MISCELLANY VMS Software Development Group As usual, DEC sent a number of its VMS Software developers to DECUS. This provides the attendees an ideal opportunity to get answers to very technical questions "straight from the horse's mouth" as well as giving the developers immediate feedback from the people using their products. Most people would agree that these people are extremely easy to talk to, and will give you straight answers (at least off-line). These people are the main reason so many people come to DECUS. This DECUS, the following VMS developers were present: o Kerbey Altman - Exec, Boot, I/O, CI/HSC/UDA o Mark Bell - Tuning, Performance o Wayne Cardoza - Memory Management o Meg Dumont - DECnet, Magtape o Steve Forgey - Performance Measurement o Peter George - DCL, Utilities o Tim Halvorsen - DECnet, DCL, SDA, Utilities, Ethernet o Dick Hustvedt - Exec, General VMS o Marty Jack - BACKUP, File System, Job Controller o Trevor Kempsell - Product Management o Peter Lieberworth - RMS o Dick Mahoney - Product Management - 35 - VMS MISCELLANY Page 4-2 VMS Software Development Group o Steve Marchesno - VMS Documentation o Kathy Morse - Memory Management, Multiprocessors o Robert Rappaport - Disk and Tape Drivers o Gerry Smith - SET/SHOW, LOGINOUT, Security o Dave Thiel - System Services and Executive o Keith Thomas - RMS VAX/VMS System Update During the VAX/VMS System Update session, the following items of information were announced: o VMS 3.1 has been released. It features support for the following items: - DMF32, DR750 - BASIC 2.0, PASCAL 2.0 - CI780 (Computer Interconnect) o VMS 3.2 will be released "shortly". It features support for the following: - TU80 Tape System (Streaming Tape Drive) - LN01 Laser Printer - LP27 High-Speed (1200 LPM) Printer o VMS 3.2 is "re-mastered." This means that customers receiving VMS for the first time will receive a 3.2 kit, not V3.0 plus V3.1 updates V3.2 update. ___ o VMS 3.2 will be available on RA60 media (205 MB disks). It will also use VMSINSTALL, a new procedure to allow easier and more standardized maintenance releases, and installation of layered products. Note: In an offline discussion, Marty Jack (VMS Development Team) mentioned that the bug in Standalone Backup will be fixed in VMS 3.2 (see VMS Problems for more information). o This session also announced the availability of VAX-11 APL, and VAX-11 Pascal (see section under Languages). DEC also announced several new packaged system based on the "RA" series disk drives and the TU80 tape drive. - 36 - VMS MISCELLANY Page 4-3 Replacing Object Libraries with Shareable Images Replacing Object Libraries with Shareable Images This user paper (VSIG pp. 333-338) by Edward Hamrick of Boeing Computer Services presents information "that will help you decide whether you should replace an object library with a shareable image library. It also describes the nuts-and-bolts of how to build a shareable image library from an object library." The shareable image is a feature which has been seldom exploited by the user community and bears further investigation. This paper certainly stimulated my interest! __ His approach of explaining, then presenting examples, followed by a "cookbook" approach makes it very easy to follow. Some of the advantages of a shareable image (as listed by Mr. Hamrick) are: o "If you change a shareable image library, you don't have to re-build application programs that use the library." o "Application programs that use shareable image libraries will link faster and take less disk storage..." Some of the disadvantages include: o "Programs that only use a few routines in a library take more virtual memory if they use a shareable image library." o "If a shareable image is to be put in a directory other than SYS$SHARE ([SYSLIB]) then a logical name must be created to point to it..." Some of the problems which may arise include: o "If a shareable image library contains FORTRAN common blocks that a user program must access, the user will get a warning message from the linker saying that PSECT attributes do not match. This is quite annoying to most users..." o "When a user links to a shareable image, all of the visible entry points (universal symbols) are included. If the user program has a routine named NUMBER, and if there is a routine named NUMBER as a universal symbol in the shareable image, then the linker will generate a warning message." o Large unitialized data areas take up disk space in the shareable image disk file. o "If an object library contains hard-coded calls to routines that the user supplies (instead of calling routines passed in argument lists) then you cannot convert it to a shareable image library. For example: - 37 - VMS MISCELLANY Page 4-4 Replacing Object Libraries with Shareable Images SUBROUTINE INTEGRATE (p1, p2, ,p3, ..., pn) ... CALL USERXX (p5, p6) will not work in a shareable image if the user supplies the USERXX routine. However, if the routine is coded as SUBROUTINE INTEGRATE (p1,p2,p3,USERXX,...,pn) ... CALL USERXX (p5, p6) then there will be no problem." The two program files which follow are abbreviated versions of those found in the handout. The first, SHRRASPLO.COM implements the LINK command needed to create the shareable image. The second, SHRRASPLO.MAR, creates the transfer vectors into the shareable image. Note that there is also a logical name SHRRASPLO which points to the file BCS:[RASPLOT]SHRRASPLO.EXE. This necessitates expanded file specs in certain places to avoid ambiguity. $ LINK/MAP=SHRRASPLO.MAP/FULL/SHARE=SHRRASPLO.EXE- RASPLOT.OLB/INCLUDE=IOMODE,RASPLOT.OLB/LIB,- DEV:[PLOT10.PROD]PLOT4014/LIB,- SYS$INPUT/OPT SHRRASGEN/SHARE,SHRRASIOF/SHARE,SHRCHARDR/SHARE GSMATCH = LEQUAL,1,0 CLUSTER = $TRANS,,,SYS$DISK:[]SHRRASPLO.OBJ PSECT = A__PRINTT ,LCL,NOSHR PSECT = CAD__ORG ,LCL,NOSHR PSECT = TKTRNX ,LCL,NOSHR "...The options file is specified as SYS$INPUT, and begins with reference to three [other] shareable images, SHRRASGEN..., SHRRASIOF..., SHRCHADR... These shareable images are referenced by other programs and can be different for different electrostatic plotters." "The next part of the options file is the GSMATCH specifier. This tells the image activator to allow changes to a shareable image without forcing application programs to be relinked." "The CLUSTER option tells the linker to load the list of transfer vectors from the file SHRRASPLO.OBJ, produced from SHRRASPLO.MAR which is discussed below..." "The remainder of the options file changes the PSECT attributes of all of the FORTRAN common blocks to LCL (from GBL) and to NOSHR from SHR. If you want a common block accessable from outside the shareable image, change LCL to GBL." - 38 - VMS MISCELLANY Page 4-5 Replacing Object Libraries with Shareable Images .TITLE SHRRASPLO Transfer Vector for RASPLOT .IDENT /1-01/ ; ; This module contains all of the transfer vectors into ; a shareable image. If you change this file and want to ; remain compatable with a previous version, you must add ; transfer vectors only at the end of the list. If you ; remove a transfer vector, replace it with a pointer to ; an error processing routine. ; .PSECT $TRANS,EXE,NOWRT,PIC,SHR,GBL .MACRO TRANS UNAME,LNAME .TRANSFER UNAME .IF NB LNAME .MASK LNAME JMP LNAME2 .IFF .MASK UNAME JMP UNAME2 .ENDC .ENDM TRANS TRANS PLOTS TRANS PLOT TRANS AXIS ... TRANS SCALE .END "This cookbook procedure assumes that you are building a shareable image library from an object library named XXXXXX and are putting it into a directory named BCS:[XXXXXX]. 1. Create a file similar to SHRRASPLO.COM named SHRXXXXXX.COM, except don't enumerate the common block attributes. 2. Create a file similar to SHRRASPLO.MAR named SHRXXXXXX.MAR, describing the entry points into the library that users can access. Assemble it with $ MACRO SHRXXXXXX.MAR 3. Link the shareable image with $ @SHRXXXXXX.COM shown above. 4. Go through the .MAP file by hand, with SEARCH, or with a program to produce a list of common blocks (or any PSECT with SHR and/or GBL attributes). Add this list to SHRXXXXXX.COM and relink... 5. Put the shareable image where it can be accessed by users, e.g. BCS:[XXXXXX]. Set your default to BCS:[XXXXXX] and type $ LIB/CREATE/SHARE XXXXXX SHRXXXXXX.EXE. (Note: This creates a dummy object library which uses shareable image instead of .OLB) - 39 - VMS MISCELLANY Page 4-6 Replacing Object Libraries with Shareable Images 6. Define the logical name SHRXXXXXX (usually in the system logical name table at startup time) to point to the shareable image with $ DEFINE/SYSTEM SHRXXXXXX BCS:[XXXXXX]SHRXXXXXX 7. Tell the users to link to BCS:[XXXXXX]XXXXXX.OLB; they will think they are linking to an object library while they are actually referring to the shareable image. 8. If you ever update the shareable image, also recreate the shareable image library by typing $ LIB/CREATE/SHARE XXXXXX SHRXXXXXX.EXE All programs that referred to the old shareable image will now refer to the new shareable image." VAXMATH - A Mathematical Software Library for the VAX There was a brief presentation given by Boland, Chow, Ford and White of the Computer Research and Applications group of Los Alamos National Laboratory. This package consists of about 800 routines (114,000 lines of code) to perform many mathematical functions. It is in the process of conversion from large mainframes to the VAX. For further information regarding this library and its availability, contact National Energy Software Center, 9700 Cass Ave., Argonne, Illinois 60439. Miscellaneous The following items fit into no specific category and were gathered from private conversations, comments/questions in various presentations, and "off-line" chats with various "unnamed developers." o Using the SYS$SETIME function also writes the ACTIVE SYSGEN parameters to the system image SYS.EXE. o Keep RX02 disk drives away from RL02 disk drives. The RL02 drives are very sensitive to magnetic fields exhibited by the RX02. o There will be some modifications to BACKUP so the entire directory structure won't be needed. Right now, backing up files in an n'th level subdirectory gives the entire structure above it. - 40 - VMS MISCELLANY Page 4-7 Miscellaneous o In the future, the Command Language Definition utility (CLD) will have hooks for validation of data on the command line such as times, file types, etc. o There is currently a bit in the FAB (FAB$V_DSBMSK) which is undocumented. It forces the use of only /SYSTEM logical name translation. Tim Halvorsen & Dick Hustvedt warned that this will probably not work in future releases! ___ o The only way to get rid of a process in MWAIT is to give it what it wants. It was acknowledged that this may be difficult to determine. The suggestion was to use SDA on the running system to examine the event flag mask to help determine what it's waiting for. (Note: The PAGESWAPPER carried an article on this topic.) o SEARCH seems to be able to find files which are corrupt, even though ANALYZE/RMS indicates everything is ok. o There was active support for implementing an RMS index key type of DATE so that indexed file references could be keyed by date. o The DIRECTORY/FULL command will show when "NO BACKUP" has been set. Failure to display this was an oversight. o From a show of hands, approximately 50% of those VAX 11/780 users present at the VAX Advanced Q&A have ECO 6 installed, while only about 10% have ECO 7. o A new reel of tape (fresh from certification) should be initialized using the command: $INITIALIZE/OVERIDE=(ACCESS,EXPIR)/DENSITY=n label to avoid searching too far into tape for density burst. This is especially important if tape was certified at 6250. Note that the privilege VOLPRO is needed to use the /OVERIDE qualifier. o There will be a program on the DECUS tape to trace a logical block number to the file which owns it. o For a VOLSET, what does one do if the first volume becomes corrupted? 1. INIT two scratch disks as a volset with the label of the old set. Swap the old second volume for the scratch second volume. Use ANAL/DISK/LIST of the index file. OR 2. DSC the good second volume to tape. Recover using RMSBCK. - 41 - VMS MISCELLANY Page 4-8 Miscellaneous o LIB$SIG_TO_RET maps signal codes to status codes. The trend is towards the use of signal instead of status codes so that more information about the error can be passed. o "SYSPRV and BYPASS are for special occasions - one is for Xmas, the other is for TG." o There was a request that the MAIL utility be able to delete a range of message numbers. - 42 - CHAPTER 5 LAYERED PRODUCT LANGUAGES There were a number of announcements in the languages area during the Fall Symposium. These included the first release of APL, the version 2.0 release of PASCAL and BASIC and the release of a Run-Time library, in VMS 3.1, for both PASCAL and BASIC. APL The VAX-11 APL is a native-mode interpreter with a built-in function editor, debugging aids including flexible error handling capabilities and debugging tools such as trace, stop and breakpoint. VAX-11 APL can access the same files used by applications written in other VAX/VMS languages because APL uses VAX-11 RMS for all file I/O. Three file types are supported: sequential, relative files and APL component files. VAX-11 APL provides a workspace limited only by the available virtual memory. Workspaces are saved as VMS files with the APL "SAVE" command and are manipulated just as any other file in a directory. Similarly, most of the capabilities of VMS are available in APL. Executing DCL commands, batch processing, APL command files, APL initialization files, VMS mailboxes and event flags are all available in the APL environment. Digital has extended the APL standard to allow more than one APL expression on a line, end of line comments, user defined FMT statements, sorts on characters as well as numbers, and others. A very important aspect of VAX-11 APL is that all of the features of APL are available without the need for a special keyboard. This is because VAX-11 APL has a full complement of mnemonic names for each APL character. - 43 - LAYERED PRODUCT LANGUAGES Page 5-2 FORTRAN FORTRAN Joel Clinkenbeard of the Technical Languages Group at DEC hosted a VAX-11 FORTRAN workshop on Wednesday morning. The first part of the talk centered on things new to Fortran in V3.0 which we covered in the Spring '82 report. The second portion was devoted to "Hints and Kinks" type of information and how to avoid "de-optimizing" the compiler. He mentioned that the "next major release" of Fortran will allow the user to define Record structure definitions as in Pascal and Basic. This will allow structures such as FABs and RABs to be made easily accessible to the Fortran programmer. There were several items of interest regarding Fortran I/O and how the compiler and Run Time Library interact. It was reiterated that FORTRAN I/O is COMPUTE BOUND. Some hints to _______ ___ __ _______ _____ speed up I/O include: o Use unformatted I/O o Use separate I/O statement for each LARGE unformatted array or character string o Use only one item per implied DO o Avoid non-linear expressions as loop control parameters in (implied or explicit) I/O DO Loops. o Avoid variable format expressions o Optimizations available for READ/WRITE internal are NOT available for ENCODE/DECODE statements o Compiler can perform additional optimization if FORMAT statements physically appear before the I/O statement ______ referencing them A handout was available to illustrate several of these cases. They further illustrate (in terms of calls to the RTL) exactly how each type of implied-DO call takes place. For example: o (A(I), I=1,10), (B(I), I=1,10) requires 2 transmitter calls o (A(I),B(I), I=1,10) requires 20 transmitter calls plus loop overhead o Single unformatted elements eliminate the need for RTL ___________ buffer - 44 - LAYERED PRODUCT LANGUAGES Page 5-3 PASCAL PASCAL The version 2.0 of PASCAL is compatible with V1 and provides increased performance characteristics. The new features include better I/O capability, separate compilation for each program or module with the capability to share declarations and definitions, a better "fit" into the VMS environment including system calls, new string manipulation capabilities, and other new features. DATATRIEVE There were a number of sessions devoted to various aspects of DATATRIEVE. The sessions which were attended included "DATATRIEVE Record Definition Workshop", "VAX-11 DATATRIEVE Technical Tutorial", and "Writing Reports with VAX-11 DATATRIEVE." Most of the information presented in these sessions falls into a "Hints & Kinks" category. As such, we will attempt to summarize the information content by topic. An exception to this was a series of comments relating to future directions for DATATRIEVE, which will be addressed separately. The "Hints & Kinks" sessions presumed a fairly extensive background in DATATRIEVE and this is reflected in the examples reported below. DATATRIEVE Hints and Kinks o FIND - there is almost NEVER a need to perform this function, and it imposes approximately a 2-5x penalty in time over simply using a record selection expression (RSE) in a statement. Instead use a statement of the form: FOR DEPOTS WITH TAG-NUMBER=NEW.XXX... o The construct "VALID IF VARIABLE=1,2,3,4,5" is equivalent to testing VARIABLE for a value of 1 or 2 or 3 or 4 or 5. o Where arithmetic operations are to be performed on variables, it is preferable to store them as COMP type data. This saves ASCII to binary conversions. An example in a record definition clause would be - 45 - LAYERED PRODUCT LANGUAGES Page 5-4 DATATRIEVE TEST-TIME USAGE COMP1 PIC 999V99 EDIT STRING IS ZZZ.ZZ o If you want to have "hidden" fields, such as date-stamps, etc. that won't be prompted for, you can make use of a GROUP variable of type FILLER. The actual variable would then be a group member. For example, 10 FILLER. 15 DATE-STAMP USAGE DATE EDIT-STRING NN-DD-YY. o The first n-characters of a key may also be used as a key with only a single key definition in the file. o When DATATRIEVE accesses a keyed file, it uses the FIRST key variable encountered. Therefore, a rearrangement of boolean expressions may result in faster execution. o Where possible, use relations (CROSS) between domains, rather than a hierarchy. o There is a subtle restriction regarding FIND/SELECT due to the fact that SELECTed records are considered stable and the fact that you can't make use of a collection until it is complete (due to the compiled nature of DATATRIEVE.) Thus in the following example, the SELECT statement is actually used against the YACHTS domain, not OWNERS. FIND YACHTS BEGIN FIND OWNERS SELECT PRINT END o In a reflexive join (i.e., CROSSing a domain with itself such as YACHTS CROSS YACHTS), it is important to note that using FOR for record selection works fine, while a FIND will not. For example: FOR A IN YACHTS CROSS B IN YACHTS OVER BUILDER PRINT A.MODEL, B.BUILDER o To find missing variables in a CROSS of two domains, use a construct similar to the following examples, where the first finds A with no B, and the second gets B with no A: - 46 - LAYERED PRODUCT LANGUAGES Page 5-5 DATATRIEVE FOR A CROSS B OVER FRIBBLE FOR A WITH NOT ANY B WITH A.FRIBBLE=B.FRIBBLE FOR B WITH NOT ANY A WITH A.FRIBBLE=B.FRIBBLE o To optimize join speed, use a key as the variable CROSSed over, specifying the keyed domain last. With all other things being equal, specify the domain with the fewest records first, since it will be used as a key for the other domains. o User functions must have PIC (Position Independent) code and should use stack for storage. o When using Callable DATATRIEVE, use "REPEAT" of "FOR port ..." instead of external DO loops. Use prompts to store data into domains. Make ports specific to tasks, don't DECLARE them. o When using Callable DATATRIEVE, NEVER put DTR$COMMAND calls inside a program loop. Use a DATATRIEVE loop instead. The more work that can be processed by each DTR$COMMAND, the better (faster!). o When using Distributed DATATRIEVE, keep in mind the following restrictions: - No WHILE, IN or ANY - No CONTAINING with a prompt - No CROSS or remote domain - Must use unique field names for each domain - There is no reprompting for validation errors when storing into remote node - No STORE or MODIFY of remote hierarchies Some hints for developing distributed domains: - Develop the application locally using local loopback - Maintain location (site) independent procedures o Date arithmetic - to compute someone's age, given their birthdate: - 47 - LAYERED PRODUCT LANGUAGES Page 5-6 DATATRIEVE DECLARE T DATE DEFAULT VALUE IS "TODAY". 10 AGE COMPUTED BY (T - BIRTHDATE)/365.25 o Remember the built-in functions. Also, let formatting take the place of arithmetic where possible. For example, given an age, compute the decade of an age: DECLARE AGE_RANGE COMPUTED BY FN$FLOOR (AGE/10) EDIT_STRING IS 90. DATATRIEVE Future Directions This session was presented Thursday afternoon by Dan Dietterich, DEC's project leader for VAX-11 DATATRIEVE. During the discussion, he broke the material into things which might be implemented in the NEAR future, the FAR future, and "Things we will never do." The implication was that the near future might be very near indeed! Near Future: o ADT Ease of Use - will allow user to "back up" in questions, to change his mind. It will also become screen-oriented and allow a dynamic display of fields already defined. o Guide Mode - will have an improved "feel" and will allow for both basic and advanced users. It will also be able to access more of the language, including reports, plots, and procedures. o HELP - will become screen-oriented -- allows the user to specify a help window size and text can be scrolled in that window. HELP ERROR will give additional info on most recent error encountered. o Documentation is reorganized with better examples and includes a documentation directory with a global index. o Performance has been enhanced in the following areas: - Plots will load faster - Startup of DATATRIEVE interpreter will be faster - READY will be faster - 48 - LAYERED PRODUCT LANGUAGES Page 5-7 DATATRIEVE - DBMS READY will be significantly faster - DBMS access will be substantially improved o There will be more capability for customization, allowing a shop to leave out features such as ADT, GUIDE, EDIT, etc. The text for the message files will be included so it can be modified if desired. Other user-visible text and default picture strings will also be modifiable. o DBMS - DATARIEVE performance will be improved. Multiple subschemas will be available through remotes. COMMIT and ROLLBACK are statements, together with ERASE. o Forms handling will be improved considerably. Currently DATATRIEVE reopens the forms library for each READY. This will be fixed. It will also be possible to have field-level access to a form through commands such as: - PUT-FORM fieldname - GET-FORM fieldname - DISPLAY-FORM formname o Callable DATATRIEVE is enhanced by allowing all SET options as bits in the DAB. User-defined keywords will be allowed, and there will be a stall-point to process them. DEC will make available their "terminal server". By calling DTR$DTR, it will handle getting data and commands from the terminal and allow the user to implement their own custom DATATRIEVE. o Extensions to relational access to eliminate duplicates from CROSS, as well as a multiple CROSS construct which would be executed in one pass: A CROSS B OVER FOO CROSS C OVER BAR CROSS D OVER FRIBBLE o Miscellaneous other improvements including: - Ability to EDIT any dictionary object - Remote domain security and password security enhancements - Functions with no values will be allowed - Edit-strings on domain tables - Staged output - output written to virtual memory first. This will allow easier formatting of complex reports. - STARTING WITH boolean expression - 49 - LAYERED PRODUCT LANGUAGES Page 5-8 DATATRIEVE - RELEASE [ALL] - like FINISH - ABORT with no message - DECLARE SYNONYM dynamically in addition to current static file definitions - SET [NO] VERIFY - applies to DTR command file - SET [NO] SEMICOLON - make ";" forced/optional - CHOICE statement (like CASE) Distant Future o Plots: add new ones, enhance old ones, make it better o Features - Exponentiation (A**B) - Pass arguments to procedures - READY *.DOMAIN - Support more data types - Handle time values in dates - Enhance MISSING VALUE - Allow multiple QUERY names - Enhance Report Writer - Improve query optimization, e.g. CROSS - Union of multiple record streams - CENTER print object - Add default mode in the domain definition (e.g. SHARED) - Allow grouping (AT TOP, AT BOTTOM, etc.) outside Report Writer "Things We Will Never Do" o Allow general abbreviations (smallest unambiguous string) - DATATRIEVE is too dependent on context of commands -- would make interpreter "horrible" o Verify or trace procedures - source is no longer available at execution o More complex statistics such as variance, etc. o Expanded Report Writer - want to keep it simple, non-procedural o VT100 plots o Very complex plots - 50 - LAYERED PRODUCT LANGUAGES Page 5-9 FMS FMS The only FMS session attended was "FMS Future Directions". During this session, it was mentioned that the features to be described were part of FMS V2.0, which was then in field test. FMS V2.0 should be available for shipment sometime after February, 1983. FMS, of course, is DEC's Forms Management _ _ System, which is used for data entry via formatted terminal _ displays. Since a basic familiarity with FMS was assumed, only future enhancements were discussed. The components of the new ______ (Native Mode) FMS include: o Forms Editor - used to layout basic screen form. Consists of multiple phases: - Form - includes use of VT1xx line-drawing instructions set for boxes, etc. - Layout - Defining location of prompts/data on screen - Assign - Specify field information - Order - Specify in what order fields are to be accessed. By default, fields are accessed in order defined. - Test - Allows forms entry to be tested without writing any code - Data - Specify named data characteristics - Exit - Save form in file o Form Driver - subroutine package to provide linkage to calling program o Form Upgrade Utility - FMS forms prior to V2.0 will NOT be ___ compatible with V2.0 utilities until processed through this utility. o Librarian - Native mode, allows Create, Insert, Replace, Extract, Delete o Forms Application Aids - can be used to produce COBOL Data Division statements, directory of libraries, develop vector module for Action Routines o Form Tester - program to display form, allow data entry & validation. o Form Language Translator - allows form to be created by description, rather than interactively at terminal. Major new capabilities include: o User-action routines - Routines written in any VMS language which are activated when data is entered into a field, or the HELP or a function key is depressed. FMS passes a - 51 - LAYERED PRODUCT LANGUAGES Page 5-10 FMS parameter (up to 80 characters) to this routine. Typical uses include range-checking,etc. o HELP routine - called before normal FMS HELP and could be used to keep track of problem areas, or to decide where additional training or documentation is needed. o New forms attributes include highlight before reading (field changes attributes when cursor enters field) and ordering. o New field attributes - zero-suppress, VAX time/date format, on-screen time & date. - 52 - CHAPTER 6 TOOLS:VNX In a session entitled "Technical Overview of VNX", Sam Feldman of Digital Software Services explained exactly what VNX is. VNX is not a language, but rather a concept. The idea is to provide many of the tools which have been made popular by UNIX* within a VMS framework. Hence, VNX is a product line which attempts to "cross" UNIX with VMS to get VNX. (In an aside, Sam Feldman mentioned that DEC now supplies and supports a product called V7M11 which is a true UNIX for PDP-11s.) VNX has as its goals: o Efficient "C" compiler o Software portability o Productive programming environment o UNIX-like features o Consistent user interface VNX does not intend to provide: o UNIX transparency o 100% UNIX compatibility. Currently there are three VNX products for VAX/VMS: o C, C Run Time Library o DEC/CMS - Code Management System - analogous to UNIX SCCS facility o DEC/MMS - Module Management System - analogous to UNIX MAKE facility _______________ * UNIX is a trademark of Bell Laboratories. - 53 - TOOLS:VNX Page 6-2 DEC/MMS - Module Management System DEC/MMS - Module Management System MMS was announced by Lee Radebaugh (Product Manager) and Karen McNamara (Group Supervisor) as a new product in the VNX family. The information presented at the session was very sparse. Lee Radebaugh was kind enough to forward a preliminary user's manual to us, and that is the basis for much of the information in this section. MMS entered field test about two weeks before the symposium and is expected to ship in the April-May '83 time frame. What DEC/MMS does "DEC/MMS acts as a command procedure to control the building of a software system efficiently. MMS determines which components in a described system have changed, and rebuilds the system by updating only those components that need updating."* The information needed for MMS to perform these functions is found in a description file created by the user. It is a collection of rules which describes how parts of your software system are built and which parts depend on other parts. The file also describes the process of building the system, i.e., the action to apply when a change is found. The general form of a dependency rule is: TARGET(S) : SOURCE(S) [!comment] ACTION LINE(S) [!comment] A typical dependency rule and its action line would be: TESTS.OBJ : TEST1.BAS,TEST2.BAS BASIC/OBJECT=TESTS TEST1TEST2 _______________ * VAX-11 DEC/MMS User's Guide AA-P119A-TE (Preliminary) - 54 - TOOLS:VNX Page 6-3 DEC/MMS - Module Management System DEC/MMS and DEC/CMS "If DEC/CMS is installed on your system, you can use DEC/MMS to access elements in CMS libraries. If a description file does not exist in your default directory, and if you have a CMS library defined, MMS attempts to retrieve the description file from the CMS library... To indicate in a dependency rule that the file (or element) being referred to is in a CMS library, put a tilde (~) immediately after the element name." DEC/CMS Beverly Schultz's VAX-11 DEC/CMS Overview session dealt mainly with CMS future directions and goals. Apparently CMS has won wide acceptance within DEC since about 800 developers are reportedly using it. DEC sees CMS (Code Management System) in multiple roles: o Organizer - keep modules together o Project monitor - what is current status of a piece of code o Sleuth - who made what modifications when o Assistant - maintain base levels, release management aid o Communications tool - between old, new, and future developers o Repository - source library Current Status o V1.0 released June 1982 o V1.1 in field test. Expected release from SDC is early 83. This version consists largely of performance enhancements. o V2.0 is in the design stage. - 55 - TOOLS:VNX Page 6-4 DEC/CMS Future Directions: CMS Version 2.0 goals: o Performance: - Internal structure changes to library directory - Improved comparison algorithm - Callable CMS o Ease of Use - Wildcarding - allow FETCH of *.*/GEN=CLASS, etc. - Improved error messages - New parsers - Added and modified commands (13) including * Delete history * Remarks in history file * Multi-file elements can be replaced with single-file elements (GROUPS) - Callable CMS - SHOW reporting enhanced including /BRIEF, /FULL, /CONTENTS (of GROUPS), /MEMBER o Reliability and Maintainability - Verification enhancements - make certain that file contents are reliable - Merge enhancements - Callable CMS - Conversion (between release levels) aids - 56 - CHAPTER 7 NETWORKING AND COMMUNICATIONS At this DECUS, Digital concentrated its network efforts in 4 areas: o DECnet Phase IV (NSIG pp. 226-238, 240-261) o Ethernet - is a reality (NSIG pp. 1-24, 25-36, 221-225, 240-261) o DECnet/SNA gateway (NSIG pp. 102-108, 246-251) o Introduction of OBSERVER network monitor (NSIG pp. 186-211) DECnet Phase IV DNA Phase IV will be supported under the major Digital operating systems -- VAX/VMS, RSX, TOPS-20, and P/OS (Professional 3xx). As expected, it will provide coexistence with Phase III nodes. Planned enhancements include: o Ethernet support -- VAX in mid-83, RSX in early-84 o Up to 1000 nodes in net (mainly for Ethernet) o Packet-switch Interconnect (X.25, X.29, X.3) bundled into DECnet o DECnet/SNA gateway o Heterogeneous command terminals o Network mail o Cross-channel piggybacking of ACK, control messages - 57 - NETWORKING AND COMMUNICATIONS Page 7-2 DECnet Phase IV o Improved flow control o Retransmission of connect-initiate messages. Ethernet Quite a number of sessions and user papers were devoted to this topic. The Network Notes handout contains some of these. Officially, the DEUNA Ethernet UNIBUS controller and related DECnet software are in field test, with first customer ship expected in mid-83. But to whet our appetites, the exhibit area contained a small VAX Ethernet. It appeared that response time and throughput were slightly better than with fully-connected 56Kb DMR11s, even on the 11/730. Ethernet performance Madhav Marathe and Bill Hawe of DEC gave a paper on Ethernet performance (NSIG pp. 25-36). They simulated a "typical" distributed university environment to see if an Ethernet LAN would provide adequate bandwidth for a large number of users while maintaining adequate response time. They concluded that the bandwidth is quite adequate. However, as pointed out by the authors, the "typical" activities did not include the use of screen editors. DEUNA Ethernet UNIBUS controller The DEUNA implements the physical and data link layers of the ISO model. The higher protocol layers are not yet firm, but are under consideration by the IEEE as proposed standard 802. Currently, multi-vendor nets can coexist but not work together. (Analogy - you can direct dial Europe, but not be able to talk to the person that answers the phone!) In addition to DECnet, DLX-type software will be available on RSX and VAX systems to support user-developed protocols. The DEUNA can also communicate via a loop test protocol for maintenance operations. Software can direct it to perform an N-party test using the Ethernet broadcast feature. If two nodes are unable to communicate, one can request specified intermediate nodes to act as go-betweens. This is a valuable diagnostic for - 58 - NETWORKING AND COMMUNICATIONS Page 7-3 Ethernet detecting marginal transmission characteristics and overextended networks. Hello/test messages also utilize the broadcast feature which reduces the amount of housekeeping traffic. Tom Ermolovich, DEC Engineering Manager, gave a very good presentation on the DEUNA/H4000 design and operation. See the Network Notes (NSIG pp. 1-24) for details. DEC Ethernet strategy DEC's initial offerings include the DEUNA controller, the H4000 transceiver, and DECnet software for VAX and RSX systems. Over the next three years, new Ethernet nodes will be introduced -- PRO-300 computers, terminal servers, routers, and gateways to SNA and X.25 networks. With the use of repeaters to overcome the 500M distance limitation, multiple Ethernet coaxial segments may be interconnected. DECnet/SNA Gateway As a big step in support of internets, DEC announced its SNA gateway for communication with IBM systems. This product should be available for VAX by mid-83 and for RSX by mid-84. To accomplish this function, the five layers of the SNA model have been distributed. The gateway implements the common network layers - path control and data link control. A DECnet host creates the half-session context - presentation services, data flow control, and transmission control. The two portions are connected via a standard DECnet link. Gateway features The gateway appears to the IBM side as physical unit type 2. It performs pacing, segmentation, event logging, and loopback testing. Capabilities provided to the user include: o Remote Job Entry (RJE) - NJE and JES2 subsystems Multiple printer, punch, and reader queues Multiple active consoles - 59 - NETWORKING AND COMMUNICATIONS Page 7-4 DECnet/SNA Gateway ASCII/EBCDIC translation Compression/decompression/decompaction Remote data sources and sinks Workstation management utility Folded into DCL: $ SUBMIT/SNA o 3270 Terminal Emulation - CICS, IMS, TSO subsystems IBM 3274 model 1C, or 3278 model 1 or 2 Multiple screen sizes Protected fields Advanced video attributes Local and host copy Folded into DCL: $ SET HOST/SNA o Application interface - task-to-task communication Provides functions of CONNECT, LISTEN, ACCEPT, REJECT, TRANSMIT, RECEIVE, TERMINATE, READEVENT High level language support Dynamic buffers Path control interface Gateway components The gateway hardware consists of a PDP-11 processor with: o Dual RX02 floppy o Console o DECnet link (typically DMR11) o SNA/SDLC link (typically DUP11) The gateway system software includes: o Operating system o DECnet o SNA communications kernel o SNA access servers o Network management The DECnet host software includes: o SNA application interface - shared image, object library, definition files for languages, help files o 3270 Terminal Emulator - utility program, keyboard frame, help file o Remote Job Entry - operator and control programs, DCL support, help file - 60 - NETWORKING AND COMMUNICATIONS Page 7-5 DECnet/SNA Gateway o Gateway management - configuration procedures, system analyzers, protocol trace utility, help files OBSERVER Network Monitor Paul Messier, Engineering Product Manager, and Gregg Thibodeau, Network Software Development Manager of DEC gave a presentation on a new DECnet real-time monitor system called OBSERVER. This is a layered software package that resides on an RSX node (M or M) and runs on PDP 11/23-PLUS, 11/24, 11/44, and 11/70 systems. The purchase of this product currently includes pre-installation consultation, an on-site orientation session, and a performance review session in addition to the usual software license, binaries, and documentation. A "future release" of OBSERVER was demonstrated in the exhibit area, monitoring its DECnet Phase IV environment. The product that is now shipping supports Phase III networks. OBSERVER Features OBSERVER performs comprehensive data collection and report generation. It detects and logs status changes in a memory-resident data base. This information is available in operator displays and historical reports. Using the standard network management protocol it: o Monitors reachability of nodes o Polls reachable nodes for - Link configuration Link connection Link status Node and link counters o Detects failures of nodes and links o Detects degradation of communication links 3 logging thresholds for each link o Detects traffic overload of nodes and links high/low logging thresholds - 61 - NETWORKING AND COMMUNICATIONS Page 7-6 OBSERVER Network Monitor o Displays overall net operation and statistics Summary status display Node detail display Statistics displays - Node summary, link traffic, link errors All qualifying events and statistics are written to "shift log" files, divided into 3 sets per day. OBSERVER supports the printing of event log files and includes 6 weekly/monthly report programs. The log files are written in RMS sequential format. Optional DATATRIEVE procedures can be used to create user-defined reports. OBSERVER maintains an on-line network data base which is divided into two parts: o Definition - monitor parameters Region definition records Node definition records o Reference - used by net operations staff Node reference records Link reference records A forms-based utility is provided for maintaining this data base. OBSERVER Limitations OBSERVER is a powerful network monitoring tool. It will detect ____ failures and provide high-level aid in their diagnosis. It can alert the operations staff to saturated link conditions and long-term traffic trends. However, its role in the overall network operation is limited - it does not perform magic and it is not intended to replace NCP. As with any tool, its effectiveness is limited by the knowledge and experience of the person using it. It is the trained operator who specifies the collection criteria and interprets the displays. OBSERVER is currently available only for an RSX host. But this does not restrict it from monitoring ANY remote DECnet node. ___ Because it uses the standard net management protocol, it is as insensitive to operating system differences as NCP. Since the PRO-300 computers are essentially RSX systems, attendees asked when OBSERVER might be supported on them. The presenters could not give a definite answer. - 62 - NETWORKING AND COMMUNICATIONS Page 7-7 Miscellaneous network topics Miscellaneous network topics o Broadband networks - even with all the fuss over baseband LANs such as Ethernet, the most prevalent, mature technology is still broadband. It provides a more comprehensive solution when there are requirements to handle audio, video, or security devices. Tom Rarich, DEC Distributed Systems Harware Product Manager, gave a good talk on broadband techniques (NSIG pp. 60-84). Harold Weber described Brown University's BRUNET broadband LAN (NSIG pp. 79-84). Several sessions dealt with tradeoffs when defining a LAN (NSIG pp. 38-58, 86-101). o Packet-switched networks - The latest versions of DEC's Packet-Switched Interconnect product are VAX-11 PSI V1.2 and RSX-11 PSI/M V2.0. Paul Messier, PSI Engineering Product Manager, and Bill Spencer, Senior Software Engineer, spoke on the topic of public data networks, X.25 specifications, and the PSI product (NSIG pp. 168-184). o Although we did not attend the DECnet-RSX sessions, we understand that Colin Strutt of DEC's Distributed Systems Development Group gave an excellent presentation entitled "A Programmer's View of RSX DECnet" (NSIG pp. 110-144). o DEC's Colorado Support Center (formerly Telephone SC) has offered DECnet support since Spring 1982. o Network Security topics are covered in the Computer Security chapter of this report. Miscellaneous tidbits from the Networks Q & A session: o When hooking a DMR11 to a DMC11, the DMC11 needs an ECO. Also, the installer must correctly set the hi/lo speed microcode switch. If the two ends are incompatible, abutted packets will fail. This occurs only under DECnet-VAX 3.0 or later, since previous versions did not support receiver pipelining. o No news from DEC on utilization of VAX buffered data path by DMC/DMR driver. o The DECnet data link drivers do not log errors (SYE). VMS "SHOW ERRORS" reports only fatal errors. o Known bug under DECnet-VAX: new log file creation fails when previous version is ;32767. - 63 - NETWORKING AND COMMUNICATIONS Page 7-8 Miscellaneous network topics o Known problems with down-line loads on multi-drop lines (DMP11). o Make sure you have adequate BYTLM, BUFIO, and ASTLM when running MONITOR via a VAX SET HOST. Symptom is "quota exceeded", process deletion. We reported a similar problem with AUTHORIZE via SET HOST. When the commands are executed from within a command procedure on the remote node, the problem does not occur. o EDT is being modified for improved performance. o Autopatch A of the RSX11M 4.0 distribution kit patches DECnet 3.0 to 3.1. o A user wished to create two "sub-nets" with no route-through between them. Suggestion was: on the common node, set the MAX ADDRESS to 1, and set the event logger to ignore event 4.7. (Not sure this will work.) - 64 - CHAPTER 8 PRO-SERIES PERSONAL COMPUTERS The PRO-300 Series is DEC's top-of-the-line family of "personal" computers. It is a menu-driven system designed for the executive or professional with little knowledge of computers. Digital has expended almost 600 man-years in the definition and development of this new line. Hardware Configuration The three versions currently available are the PRO-300, 325, and 350. The PRO-350 includes: o F11 microprocessor (22-bit address space) o RX50 - dual 5 1/4" floppy disk drive (800 Kbytes total) o RD50 - 5 Mbyte Winchester disk o 256K bytes RAM memory o Time of day clock o Communications port o Serial printer port o Keyboard o Black & white monitor, bit map video NOTE In the session on P/OS Graphics, it was stated that the RD50 5 Mbyte Winchester was a "mandatory option" to be able to use P/OS Version 1.0. This statement was - 65 - PRO-SERIES PERSONAL COMPUTERS Page 8-2 Hardware Configuration repeated in later sessions as well. According to our local DEC salesman, there will be another version of P/OS which will be floppy-based and will not require the Winchester. Expected availability of this version of P/OS is estimated as late first or second quarter 1983. The exact details of how P/OS will be split are not clear at this time, since it currently requires 1Mbytes without any user applications (and the two 5 1/4" floppies total only 800KB). Current or soon to be available options include: o Floating Point Accelerator o Extended bit map option (required for VT125 emulation) o Color monitor o Telephone Management System (TMS) o 256 Kbyte expansion memory Printers currently supported include: o LQP02 Daisy Wheel o LA100-PC o LA50 P/OS and Graphics Bill Feldman and Bill Clements of Digital elaborated further on P/OS and its graphics capabilities on Tuesday afternoon. DEC has designed the human interface to the PRO series to be menu-driven and controlled totally by function keys. Thus, the user has keys such as DO, INTERRUPT, RESUME, HELP, MAIN SCREEN, etc. permanently defined on the PRO's keyboard. The menu system consists of a tree of menus which include: o Applications o File Services o Disk Services - 66 - PRO-SERIES PERSONAL COMPUTERS Page 8-3 P/OS and Graphics o Setup o Message/Status Board o Additional applications User-defined applications are placed into the existing menu, or a new menu is formed. Presently, there are five slots in the main menu, plus 11 additional application groups with 12 slots each. Print services for the PRO include: o Start, Pause, Stop, Resume print o Setup o Print Screen - dumps bit map if present to LA100, LA50 The graphics portion of P/OS consists of hardware support for 1 plane of 960 by 240 pixels. This can be augmented by the Extended Bit Map Option, which supports two additional bit planes plus 8 of 256 colors. The graphics support library is based on the SIGGRAPH core standard and is clustered and shareable. It includes: o Support for 2-D graphics o Coordinate transformation including window, viewports, origin and clipping o Output primitives for position, markets (Dow Jones presentation), lines, curves and arcs, and text PRO-300 Series Communications and TMS The serial interface port of the PRO-300 series supports both RS-232 and RS-423. It features full or half duplex communication at rates to 19.2 Kbaud with full modem control and has CRC for synchronous. Access to the communications features is provided through a Communications Service Library which allows a high-level (no QIO needed) interface. The Communications Port Driver handles the asynchronous mode I/O while a Telephone Management Driver (to be available in the NEXT release of P/OS) supports the Telephone Management System. - 67 - PRO-SERIES PERSONAL COMPUTERS Page 8-4 PRO-300 Series Communications and TMS The modem control features auto-answer and auto-dial, with settable line characteristics (including separate transmit and receive speeds). The secondary communications port can be treated like a second terminal port. On input, the user can select to read a fixed number of characters, read with timeout, etc. - no input terminator flags are used. The PRO includes emulation of two DEC terminals, the VT102 and the VT125. If you have the Extended Bit Map option, the system automatically configures the terminal as a VT125. It was noted that there were hardware differences between the PRO and the VT125 but that "most of the REGIS graphics will work -- and the differences are documented." The file transfer package consists of two parts, one that runs on the PRO, and a cooperating process running on the host. Host packages are available for RSX11M, RSX11M, and VMS. Transfer rate is approximately 1 block/second at 9600 baud. It is possible to perform file transfer in a background mode, if you have extended memory. The Telephone Management System (TMS) is one of the most exciting features of the PRO and will be available with the next release of P/OS. The hardware supports two phone lines at either 75-300 baud (Bell 103) or 1200 baud (Bell 212A) with options for multiline phones. The system includes the capability to use autodial on either line 1 or line 2 and to talk on one line while transmitting data on the other. A CODEC (COder/DECoder) allows voice digitizing via Continuously Varying Slope Delta Modulation at 32 Kbits/second (this translates to about 8 disk blocks/second). Typical applications include voice mail, voice response, dictation with voice editing, telemarketing, etc. A product manager described a talking "slide" program he had developed which blended fancy graphics displays with digitized voice. The TMS driver can also decode touch tone signals and thus could be used for slow data collection applications. An accessory voice unit includes a keypad, microphone/speaker, and various other controls. Software Software for the PRO series falls into three general categories: 1. DEC-supplied software which runs on the PRO 2. DEC-supplied development tools which run on a host connected to the PRO - 68 - PRO-SERIES PERSONAL COMPUTERS Page 8-5 Software 3. Third-party supplied application packages which run on the PRO P/OS - Professional Operating System P/OS, the operating system for the PRO series computers, is an outgrowth of RSX11M. P/OS shares a common source with RSX11M, which means that it will track fixes, upgrades, (and bugs!) as found in 11M. (Note that no means has yet been decided upon to supply software updates for P/OS.) Specific modifications are discussed in this section. o New Features - Incorporates RMS-11 V2.0 Named directories (but no subdirectories) Decimal version numbers Wild carding of filenames and directories Logical names for devices Random access, fixed-length files Segmented, clusterable libraries to conserve memory RMS-11 V1.8 support of indexed, relative, sequential, and block access Occupies 16KB of 64KB task virtual address space (but is clusterable) - Includes automatic mount/dismount of media - DMO removes installed tasks - Enhanced higher-level language interface to O/S and utilities - Vectored libraries - Time of day clock o Features removed from RSX11M - Virtual terminals - System reconfiguration (CON,HRC) - Batch, Queue manager, Accounting - Error logging, volume shadowing - Prototype TCB, external task headers - MCR, alternate CLI support - TDX, TKTN, VMR, HELLO, BYE utilities - Indirect command procedures (IND) - on wish list - File Control Services (FCS) - Group event flags - Checkpoint file - 69 - PRO-SERIES PERSONAL COMPUTERS Page 8-6 Software PRO Applications Development Tool Kit Since the only "development" software which actually runs on the PRO (at this time) is the PROSE editor and PRO-BASIC, any application development must be performed on a host (RSX, VMS) system and the task images transferred to the PRO for debugging. This is admittedly a cumbersome process and one can expect to see the PRO hosting other languages in the future. (As one of the developers remarked, "We're sick of it too. It's one of OUR highest priorities!") Languages now available include BASIC2, MACRO-11, and DIBOL with FORTRAN '77, COBOL '81, PASCAL and C coming. The development tool kit includes: o Frame Development Tool (FDT) which is interactive and creates message, help and menu files. o Forms Management System (FMS) which creates forms for data entry. The forms editor (FED) runs on the host, while the forms driver (FDV) is linked into the PRO application. o SORT - a record or tag sort package which is also accessible via subroutine calls o Debugger for BASIC2, DIBOL, etc. which can make use of a second terminal through the communications port for debug messages. o Communications support o File transfer package Available separately, and at extra cost are the languages mentioned above. One must also buy the host end of the file transfer package separately from the tool kit. NOTE Prices and details for the PRO are contained in the PDP-11 Options & Summary book. However, it has been verified that the statements shown there which imply that the languages are included as part of the price of the Development Tool Kit are INCORRECT. Each language must be purchased individually. - 70 - PRO-SERIES PERSONAL COMPUTERS Page 8-7 Software PRO-BASIC PRO-BASIC is a language interpreter running on the PRO as a layered, P/OS application. It is ANSI compatible and is a subset of the normal DEC BASIC. It includes support for sequential files, virtual arrays, IF-THEN-ELSE and error handling. It does not provide ISAM file capability. Extensions include a CHAIN command with parameters, CALL COLLATE for international character set comparisons, interactive debugging and color graphics. UCSD P-System It was announced that it is possible to buy the P-System for the P/OS from Softech. It is based on UCSD V4.1 and acts as a layered product on top of P/OS. It will be distributed by DEC through retail outlets and consists of two packages -- a development environment (compilers, linkers, editor, etc. which run on a "virtual disk"), and a run-time only environment. Note that once inside the P-System, there is no access to any of the standard P/OS system calls. It is possible to get PASCAL, FORTRAN and other languages for the P-System. PRO-300 Future Directions This session, presented by Don Jenkins of DEC, talked of the future possibilities for the PRO line of personal computers. As such, it must be taken as a "wish list" and not as a firm commitment from DEC. That is, some things are much more likely to occur than others. Areas where one may expect to see changes include: o Higher capacity disks, especially hard disks to give more storage and backup capability o Personal computer clusters, possibly using Ethernet for 6-8 machines o DECnet/Professional using gateways to link to other systems o Higher performance - 71 - PRO-SERIES PERSONAL COMPUTERS Page 8-8 PRO-300 Future Directions - J-11 32-bit microprocessor - Memory expansion - boards are laid out for 256K RAM chips when available - Higher density Winchester drives - possibly to 25-50 MB o Increased video resolution in graphics mode -- from 1024x256 (now) up to 1024x512. Other features might include multiple windows, new input devices such as mouse or tablet, integration with video disk. o Additional software tools: - Simple database/query - Graphics editor - Improved debugging aids - Character set editor o New communications support: - TELETEX for Europe - IBM interconnect (via gateways) COMDEX Product Announcements DEC announced a wide variety of software and hardware products at the COMDEX trade show held recently. These included: o DEC Software - PROSE - closer to word procesor editor, a superset of PROSE. Also gives document merge, justification, cut/paste, etc. Available May. - MACRO, BASIC, DIBOL2 - available in toolkit in January '83 - PASCAL, COBOL-81 - available in toolkit Spring '83 o Hardware - CP/M option board with Z80A, 64KB RAM, ROM, Bootstrap. It runs CP/M-80 and applications tasks and is compatible with Rainbow & DECmate. CP/M is invoked from the P/OS menu, then boots CP/M and runs from virtual disk and floppy. Available July '83 - 256 KB RAM (Max of 3 to give 1MB) - Available Spring '83 - Upgrade PRO-325 to 350 - 72 - PRO-SERIES PERSONAL COMPUTERS Page 8-9 COMDEX Product Announcements o Third-Party Software Available through DEC - IBM 3276 Bisync emulator - Executive Desk Set - Executive Data Management - data presentation/report - LOGO - MUMPS/PC300 - PALETTE - Interactive 2-D drafting - SPSS-11 Statistics Package - TK SOLVER for equations, graphs - Universal Microprocessor Development - Pascal, disassembler, etc. - VISIFILE - Forms-oriented filing system - Advanced VISICALC - VISIon Furniture The Accessories & Supplies Group exhibited its new line of modular furniture tailored for DEC's personal computer family. Components include a workstation, terminal table, portable pedestal file, and numerous accessories. The workstation incorporates a sexy feature - a motorized post holds the PC wedge monitor, permitting it to be positioned and swiveled for maximum user comfort. These furniture products will be available in March 1983. See the January '83 DECdirect catalog for details. - 73 - CHAPTER 9 HARDWARE MIL-Spec Hardware Gary Miller of the Government Systems Group gave a presentation on DEC's philosophy regarding MIL-spec hardware. Basically, they feel that there is not enough of a market to justify their heavy involvement directly in this area. Instead, they prefer to license other manufacturers to build MIL-spec versions of their products. Examples of this include the Norden PDP-11/44, and the Norden MIL-spec VAX 11/750. Raytheon is also being licensed to build a special version of a computer with a "subset of the VAX instruction set architecture" for special DoD programs. DEC does make several versions of RFI/EMI cabinets and talked about future fibre optic interfaces to the DMF32. TEMPEST DEC no longer manufactures a TEMPEST terminal. Instead, they have signed a contract with Delta Data (a Philadelphia company) which will OEM VT100 components, manufacture a TEMPEST housing and perform integration and testing. They are also currently discussing doing the same thing for the VT125. They acknowledged that there was a "chance" that the PRO-350 might be adapted for TEMPEST requirements, and noted that all of their personal computers are in compliance with FCC Class-A regulations for emissions and interference. - 74 - HARDWARE Page 9-2 VAXStation-100 (VS100) VAXStation-100 (VS100) The VAXStation family of products was announced at the Spring '82 DECUS. The current member of this family is the VAXStation-100 and will be the subject of this discussion. Components The present configuration includes a 19" B&W monitor, keyboard (same as PRO line of personal computers), an optional tablet and mouse, and a graphics processor. It has a fibre-optic interface which supports a link length up to 1000 feet to its Unibus interface on a VAX host computer. A color system is coming, but not here yet. The graphics processor consists of an M68000, four 2901 bit-slice AUs as a vector accelerator, and 4 pages of display memory, each consisting of 1 bit by 1024 x 1024 pixels. Architecture The VAXStation's System Display Architecture (SDA) can handle 4 separate windows (processes) simultaneously, and display the information from the window anywhere the user desires. The important goal to be achieved was the separation of processing and presentation. Thus the application writes to a Virtual _______ Display. The attributes of the virtual display include: text, _______ graphics, size, foreground and background color. The user/application then composes images using a Pasteboard where __________ virtual images are combined. The user, in turn sees a Window ______ through a Viewport on the physical screen with the following ________ effects: Move Image, Magnify or Reduce, Zoom In or Out, Pan, Clip/Expose. All data is maintained by the system, regardless of visibility. To the user, there appear to be separate text and graphics planes, where the graphics image is stored as a bitmap. Software tradeoffs were made in favor of higher display performance (writing to the screen). A block diagram of the Software Architecture is shown below. - 75 - HARDWARE Page 9-3 VAXStation-100 (VS100) ------ -------------- --------- ---------- -------- | | | Program | | |-->| ACP |-->|VS100 | |Appl|-->| Library |-->|Virtual| ---------- | | | | -------------- | | ^ |Device| | | -------------- | | | | | |Prog| |Professional| |Display| ------- |Driver| | |-->|Graphics Lib|-->| | / \ | | | | -------------- | | / Human \ | | | | |Driver | \Interface/<--| | | |------------------->| | \ / -------- ------ --------- -------\ \---Mouse Features & Availability The VAXStation-100 features emulation of the DEC VT102, the Tektronix 4014, as well as its own "native text" and "native graphics" modes. DEC has a "well-defined protocol" for new emulators, which should make it easy to add new versions as needed. The VT125 is currently NOT emulated but it will ___ "probably be in the 2nd release of the software". There are no separate tools provided to help the user write his own emulators. No local hardcopy is available, but it is expected to interface with the new LN01 Laser Printer (described later) through the host. The graphics library is based on the SIGGRAPH CORE standard. There are no present plans to implement the GKS (Graphical Kernel Standard). Since the VAXStation can support multiple processes and places a high load on the host, the following configuration guidelines are currently given: o VAX 730 - 2 VS100 o VAX 750 - 4 VS100 o VAX 780 - 6 VS100 A Technical Summary will be available in January or February '83. The VS100 is expected to be available for shipment by June '83 and will probably cost about $10,500. The presentation was given by Bruce Bayuk (Product Mgr.), Daniel Ganek (S/W Architecture), and John Freitas (Layered S/W). - 76 - HARDWARE Page 9-4 LN01 Laser Printer LN01 Laser Printer The LN01 is DEC's entry into the field of laser printing. It is a small, relatively slow (by laser printing standards) system. The LN01 prints one page at a time, twelve pages per minute (about 600 lines/minute). It uses a font which allows it to print up to 132 columns at 13.6 characters/inch and 8.57 line/inch on one side of the paper only. It has a resolution of 300x300 dots/inch and prints on 8.5x11 or 8.5x14 inch cut sheet paper. A DEC spokesman jovially mentioned that reliability is measured in MTBPJ (mean time between paper jams). Unfortunately, it is not suitable for plotting applications, since it does not work on a line-at-a-time basis (as does an electrostatic printer or the LXY21 (Printronix) printer). Its principal market is the photo-typesetting industry and where very high quality output is required. The LN01 is rated for 10,000-15,000 pages/month. DEC appears to take this seriously since the monthly maintenance charge is $320 plus 2.6 cents/page in excess of 15,000. LP27 High Speed Printer The LP27 is a high-speed band printer capable of printing "a minimum of 1200 LPM using a 64 character print band and 800 lines per minute using a 96 character print band. Both bands are included."* The 64/96 character bands refer to all upper case and upper/lower case, respectively. The printer is made by a third party to DEC's specifications. Standard features include a paper puller and dual sets of form feed tractors. DEC claims that "ribbon, print band or paper change take less than three minutes."* _______________ * LP27 Brochure YM-AS37A-00 - 77 - HARDWARE Page 9-5 LA100 Letterwriter/Letterprinter LA100 Letterwriter/Letterprinter New fonts for the LA100 will likely be available from A&SG; candidates are APL, italic, and scientific. Details on customer-definable fonts will be announced in Jan-Feb 83. New Mass Storage Architecture/VMS Support There were two sessions to discuss DEC's latest mass storage devices and their interface with VAX and VMS (VSIG pp. 381-391). DEC has introduced a new line of disks, tapes, and interfaces which are far superior in price, performance, and longevity to all their current devices -- MASSBUS and UNIBUS alike. The disk devices listed all interface to the UDA50 (available now and supported under VMS Version 3.1) and to the HSC50 which was just announced (see below) and which will be supported "in a future major release of VMS." Devices The devices announced include the following: o UDA-50 - first controller implementation of the DSA. It will connect up to four disks via the UNIBUS. The UDA50 is 2 hex size boards. It supports Winchester fixed-media disks (RA80,RA81) as well as removable-media disks (RA60). The UDA50 operations are controlled by a high-speed, 16 bit processor which simultaneously executes host interface and drive interface programs. The UDA50 firmware manages disk accesses, data transfers, performance optimizations and expanded error recovery. The controller also uses device independent software through the Mass Storage Control Protocol (MSCP). Disk-dependent requirements are isolated form the host operating system by the MSCP device driver allowing different types of disks to be connected to the UDA50 without operating system modifications. Features of the UDA50 include: - Command Queue - stores up to 12 data transfer requests (controlled by a SYSGEN parameter, UDABURSTRATE). Firmware controlled operations on the command queue manage disk accesses and data transfers. - Seek Ordering - The UDA50 sets up logical command queues for each disk. It can rearrange and execute requests by cylinder address rather than in the order - 78 - HARDWARE Page 9-6 New Mass Storage Architecture/VMS Support received. This elevator algorithm reduces seek distances for minimized seek latency. If required, an express queue is available for FIFO service requests. - Overlapped Seeks- The UDA50 initiates simultaneous seeks to all disks with I/O requests pending. This reduces seek time and allows one disk to transfer data concurrently with seek operations on other disks. - Rotational Optimization - When more than one disk is postioned on-cylinder, the UDA50 selects the drive nearest its requested sector to perform data transfer. - Speed Matching Buffer - With the 12 sectors of RAM in the UDA50 the disks can perform disk-to-controller transfers at a higher peak rate than controller-to-host transfers. The UDA50's suspend/resume capability manages transfers that exceed the buffer size without any host involvement. The UDA50 provides extensive error detection and correction. It uses an expanded Error Correction Code (ECC) which can detect and correct up to eight independent error bursts, each of up to ten bits in length occurring anywhere in the sector. An Error Detecting Code (EDC) is appended to each sector to permit checking controller memory errors. The UDA50 records each sector ID address in quadruplicate to ensure that the loss of up to two copies of sector addresses do not render the sector unreadable. The UDA50 contains microprocessor controlled fault isolating diagnostics. This makes fault detection and isolation independent of the host processor. Functional tests are run automatically during power up and initialization. Due to the seek optimization, adding a second drive to a UDA50 configuration will provide an increase in I/O throughput. However, the addition of a third or fourth drive will not yield much more improvement. Placing two UDA50s, each with a single drive, on the same UNIBUS will result in a slight degradation in the I/O throughput over two UDA50s each on a seaparate UNIBUS. If two UDA50s are to be configured on the same UNIBUS it is recommended that they be the only devices on the bus. In this configuration each UDA50 waits 1.5 microseconds after transfer to allow the other controller to gain access to the bus. o HSC50 - Gives disks and tapes access to multiple hosts - Enhanced throughput optimization by buffering multiple requests - Overlaps error handling with I/O - Provides volume shadowing capability (redundant recording of information on different disk drives) - Features online diagnostics - 79 - HARDWARE Page 9-7 New Mass Storage Architecture/VMS Support - Provides for modular expansion of devices and functions - Has controller utilities built in such as disk to tape backup, formatting, etc. - Currently supported by DECSYSTEM-2060 software - VMS support in "a future major release" o RA81 - Fixed Media Winchester disk. The RA81 rack-mountable disk subsystem is the newest of the just announced UNIBUS disk devices utilizing the UDA50 intelligent controller. The RA81 is Winchester technology with the capacity of 456 Megabytes of storage space per drive. The drive mounts in a standard 40 inch H9642 series cabinet with up to three drives per cabinet for a total storage capacity of 1.4 billion bytes. The performance data provided for the RA81 is as follows: - 6 millisecond track-to-track seek time - 28 millisecond average seek time - 36 millisecond average access time - 2.2 Megabytes per second data transfer rate - 456 Megabytes formatted capacity per drive The drive features: - An advanced read/write system where a unique encoding/decoding scheme called 2/3 rate tree code is used. - A dual positioner system where two mechanisms are used to postion the read/write heads. - The Winchester technology with the enclosed disk head assembly protects the recording head and surface. - Dual read/write heads address each data surface reducing seek activity and shortening average seek distance, thus improving performance. - Dual microprocessors - One to communicate with the UDA50 and one to perform the real-time operations allowing simultaneous tasks to occur. The RA81 is fully compatible with the previously announced RA80 drive in a mixed drive subsystem with a single UDA50 controller. The speed of the RA81 is comparable to that of the RA80. The one caveat expressed was that the speed of the RA81 on a VAX 11/730 is 5% to 7% slower than on the 750 or 780 due to the lack of certain buffers in the 730. o RA60 - Removable Media disk - 205 Mbytes/drive - Up to 615 Mbytes (3 drives) per cabinet - Up to 820 Mbytes (4 drives) per HSC/UDA50 data channel o RA80 - Fixed Media Winchester disk - 121 Mbytes/drive - 80 - HARDWARE Page 9-8 New Mass Storage Architecture/VMS Support o TA78 - Dual Density Magnetic Tape (Master) - 6250/1600 BPI 125ips Read/Write - TS78 microprocessor controlled formatter/controller - 781 Kbytes/sec max transfer rate (6250 GCR @ 125 ips) - Uses TU78 drives as slaves. - TU78s can be field upgraded to TA78 by changing formatter, bulkhead, front panel. - Has RS232 diagnostic port for terminal or handheld controller - TA78 can back up RA81 on 4 reels in less than 1 hour. - maximum capacity is 145 Mbytes/2400-ft reel (8KB blocks) o TU80 Magnetic Tape Subsystem - At the opening of the symposium Digital announced their new low cost streaming tape subsystem, the TU80. The TU80 is a half inch, 1600 BPI, phase encoded, 25/100 IPS magnetic tape subsystem. The TU80 employs start/stop and 9 track streaming tape technology. The TU80 has three speed operation optimized by the TU80 controler with no operator/user intervention. These include 25 IPS-start/stop, 25 IPS-streaming and 100 IPS-streaming. The TU80 mounts horizontally in the same cabinet as an RA81 or RA80 and in fact was shown in the exhibit area mounted with an RA81 drive in the bottom of the cabinet. One of the major attractions of the TU80 is its mechanical simplicity which should create a more reliable drive. The subsytem includes extensive diagnostics to isolate TU80 failures and performs a power-on health check each time the TU80 is powered on. One caution noted was that the effective speed (inches/second) from the TU80 depends on the speed of the processor and the rate of data delivered to the drive. Some benchmarks running BACKUP were given as - 11/780 90 IPS - 11/750 70 IPS - 11/730 25 IPS In order to keep the TU80 streaming, data must be delivered at a high rate of speed. For this reason, the DCL command COPY is not recommended for use on the TU80. COPY does too much shuffling of buffers to keep the TU80 streaming. The larger the size of the file the better the TU80 can work with COPY. BACKUP, on the other hand, does a nice job of keeping the TU80 streaming. To help in this endeavor a new qualifier will be added to VMS 3.2, /BUFFER. This will specify the number of data buffers available to BACKUP. Suggested value was /BUFFER=5. - 81 - HARDWARE Page 9-9 New Mass Storage Architecture/VMS Support A benchmark of elapsed time to BACKUP a full RA80 disk drive onto a TU80 was given with the different processors. Of course, file advancing and rewind will occur at 100 IPS on the TU80. 11/730 11/750 11/780 TU80 80 min 40 min 30 min TU77 N/A 28 min 17 min Software Support To support these devices, DEC has introduced a new Mass Storage Architecture which consists of the following components: o New DEC Standard Disk Format (DEC Std. 166) - 18 bits - Hosts see NO bad blocks - Some sectors in each cylinder reserved for replacing bad block - Host sees media in terms of Logical Block Numbers (LBN) 0 to N-1, with all mapping of LBNs to Physical Blocks handled totally by the controller. o Mass Storage Control Protocol (MSCP) - High level message syntax - Hosts request I/O using MSCP message format which specify who wants what by specifying unit numbers, opcodes, modifiers, byte count, LBN, buffer descriptor, etc. - Controllers report results via MSCP End Messages which are echos of the request plus a status result o Systems Communications Architecture (SCA) - Communications protocol between cooperating system fork processes in a cluster environment - VMS implementation of SCA is known as Systems Communication Services (SCS) - Processes are known as SYSAPs (System Applications) and include Disk Class Drivers executing in the host, and MSCP Servers executing in the controller. - 82 - HARDWARE Page 9-10 Terminals Terminals In the terminals and printers panel a few tidbits were heard: o DEC has "fessed up" to a number of known bugs in the VT100. Two that we have experienced are - Entering the reset state when in 132 column mode Interactions between jump and smooth scroll These and other bugs were fixed in the later designs - VT101, VT102, VT131. There are no plans to fix the VT100/VT125. The engineering cost is too high and most customers are buying the newer, less-expensive versions. o No plans for block mode in future terminals. o All DEC terminals momentarily drop DTR when in setup mode. o For amusement, try sending the following escape sequences to a VT125: [145q - will sound bell continuously [155q - buzzes when you press keys UNIBUS Timing Considerations Ken O'Mohundro of Able Computer, and Art Hermes of Seismograph Services Corp. provided advice on interfacing to a UNIBUS: o "When is a UNIBUS not a UNIBUS?" The PDP11/34 was the beginning of a changing UNIBUS specification. Previously, the data was guaranteed valid until MSYN became inactive. Now, the master may remove data as soon as it see SSYN. Moral: Don't lie when you assert SSYN. Address, data, and control signals must be stable for 150 ns prior to asserting MSYN. o The 20 usec SSYN timeout spec is violated - DECsystem 10/20: as low as 5 usec VAX 11/750: ~7 usec Most others: ~10 usec o In the VAX family, the coupling between the CPU and the UNIBUS varies: 11/780 Interrupts & DMA loosely coupled 11/750 Interrupts tightly coupled, DMA loosely coupled 11/730 Interrupts & DMA tightly coupled - 83 - HARDWARE Page 9-11 o According to Art Hermes, the 11/780 UBA violates the rule of the CPU being the lowest priority UNIBUS requester. This is because the UBA is really the highest priority requester: v-------v-------v-------v------< Arbiter | | | | Term D1 D2 UBA | SBI If the UBA microcode is too slow, the BR vector can be "missed" (=0). This shows up as an "unexpected UBA interrupt" in the error log. (We have noticed this problem with the DR11-B {and one look-alike} when setting its ATTN signal.) o The 11/780 UBA still has a problem with byte-oriented transfers via the buffered data path (BDP). When a UNIBUS device requests the 7th byte, a pre-fetch is initiated. The 8th byte will be from the next quadword! To avoid this problem, DEC's LPA11 driver uses longword-aligned access (LWAE). o The critical UNIBUS arbitration signals are NPR, NPG, BR, BG, SACK, and BBSY. o "NPR monitoring" locks out interrupt and CPU requests and can cause data lates. It will not work at all on a VAX 11/780 UBA. "SACK monitoring" is poor, and if there is more than one device of this type, it is worse. o Don't use "hog mode" for NPR. Holding the bus for 2-4 words is OK IFF you have all the words available. ___ o Using a uP to emulate hardware seems clever, but it make take more time on the bus. A uP design is more effective than a uP emulation. Because of the low SSYN timeout value, keep uP response time to under 5 usec. o The maximum bus length for the 11/780 SBI is 3 meters. o The 11/780 console interface board feeds power to the floppy. If you observe the flopply wildly clunking, this is the likely cause. o BA11-K chassis will now accept 32A power supplies. - 84 - HARDWARE Page 9-12 Power Problems Power Problems In the power problems workshop, attendees heard presentations from three parties. Major points were as follows: o Bill Coleman of DEC - Tap switching regulators work horribly with non-linear loads (such as computer power supplies). They tend to distort a sine wave into a square wave. - Ferro-resonant regulators need "headroom" to prevent voltage sags to as low as 40V on inrush currents (such as disk drive startup). - DEC systems typically draw high neutral current because of high harmonics, even if the phases are well balanced. - All DEC power supplies are rated in the U.S. at 120V +7%/-13% (104V-128V). - The DEC power distribution system (PDS) accepts external alarms such as water, smoke, temperature, etc. It can be set for alarm only, shutdown, or delayed shutdown. - We pointed out that we have rejected the PDS because it remains off when power is restored after an outage. This defeats the VAX auto-restart feature. DEC is still considering making this a selectable "feature". - When doing facilities planning, refer to the site preparation guide - part number is COR-SPP-003. o Ken Brill of Atlas Energy Systems - A paper by Allen and Siegal of IBM titled "Monitoring of Power Line Disturbances" is still the best collection of data on this topic. - Power transients damage integrated circuits which do not fail until they are later stressed by power and temperature cycling. - The west coast phenomenom called "flicker and flash" occurs every year because of salt buildup on high voltage insulators. - Power-factor correction capacitors are switched in and out during the day by the utility company causing nasty transients. - Isolation transformers are for common-mode noise only (line-neutral). - DEC's PDS provides isolation and a single-point ground, but no energy storage. - Uninterruptible power supplies are for power continuity. But the problem is power quality. His recommendation is a motor-generator, which Atlas makes. The Atlas Powerpod is a motor-generator and PDS. Maintenance is low -- replace bearings every 3-5 years. The noise level is 70-75 dB, even though the case is - 85 - HARDWARE Page 9-13 Power Problems soundproofed, so place it accordingly. It also gives off a good deal of heat. o Al Muir of Computer Power Solutions, Houston, Texas - A Metal Oxide Varistor (MOV) is a low-cost spike clamper. But it is usually installed incorrectly across line and neutral. The correct method is to install two - from line to ground and from neutral to ground, so that energy will be conducted to earth ground. - RFI/EMI filters must be run at near capacity to be effective. This is because of their non-linear relationship between rated/actual current and effectiveness. - When you are specifying an isolation transformer for a facility, always tell the electrician you need a shielded isolation transformer. Few electricians ________ understand the difference. - Ferro-resonant transformers are good for single phase loads of <= 5 KVA. Switching power supplies reflect harmonics back to the ferro and distort the waveform. - Isolated ground receptacles eliminated 80% of the problems at a Bell site. - IBM won't sell a large system without a motor-generator. - A "Dranetz-type" power line monitor is highly overused and overrated. It doesn't see short duration transients. - Uninterruptible power supplies (UPS) are useless unless they supply the air conditioning as well. Also, OSHA installation regulations add 50% to the cost. - 86 - CHAPTER 10 RSX/VMS RELATED TOPICS AME - Applications Migration Executive The RSX Development Group gave a presentation on the future plans for the VMS AME. (The RSX Group "owns" VAX compatibility-mode software.) Current AME Status Relative to RSX o AME is still RSX11M 3.2 o Compilers, utilities, and engineering programs migrate OK. o Problems occur for intertask communication, activation of a task by a task, and use of fast floating point. o Priviliged tasks, drivers, I/D space, and supervisor-mode libraries won't work. o MCR has syntax differences and many commands are not supported. o DECnet 11M NETGEN and SYSGEN are not supported by IND. o User pointed out difference in handling of break key in terminal driver. RSX returns error status of IE.FME (framing error). AME returns a null character and no error status. Under VMS 1.x, the AME behaved as RSX does now. - 87 - RSX/VMS RELATED TOPICS Page 10-2 AME - Applications Migration Executive Future plans for AME Compatibility mode will become a layered product, bundled with the next major release of VMS. This will allow it to be synchronized with future RSX releases. AME goals include: - Intertask communication - PLAS, M+ support - DECnet interface - I/D space - Installed tasks - IND - use M+ non-privileged ICP Track M+ directives M+ SYSGEN (next VMS release) - Add utilities (trying for BRU by next VMS release) - Track RMS-11 Currently there are no plans for: __ _____ - Connect-to-interrupt - Alternate CLI - File sharing - Shared libraries RSX -> VAX: Programming for Transportability Jim Downward of KMS Fusion shared his migration experiences and advice: o This topic usually comes up AFTER the VAX is ordered. o Don't make major migration plans until you - Know RSX and VMS very well Have read manuals and internals Have taken courses o If doing on-going development, intertask communication, or using RMS, don't use the AME. _____ o Fortran 77 code is easy to migrate; Fortran IV is harder. Differences between RSX and VMS F77 include - CHARACTER data is passed differently. Operations on LOGICAL variables do not produce overflow. - 88 - RSX/VMS RELATED TOPICS Page 10-3 RSX -> VAX: Programming for Transportability o Migration difficulty is proportional to the number of RSX executive directives. o Caveat - VMS AME clears event flags at process creation. RSX does this at task activation. o The AME gives no access to the I/O page. o Unsolicited ASTs do not work properly in the AME. o Be aware that VMS analogs may impose performance penalties - RSX SEND/RECEIVE vs VMS mailbox RSX SPAWN vs VMS SPAWN/CREPRC o Use of VMS features may improve overall performance - RSX database ACP vs VMS shared global section RMS record locking VMS lock manager o All applications are not equally "ready" for migration - poor documentation, idiosyncrasies. o Plan the application maintenance environment. It is likely that there are beaucoup RSX command files with references to UICs, etc. o Test the waters - first get very low-level routines working solidly. o Create an RSX to VMS Executive Mapping Library. Include one subroutine which will identify the O/S to mainline code to resolve differences. (Note: Jim has submitted such a library to the VMS SIG tape.) Mathematics of RSX11M This session was misnamed. The content was really advanced RSX tips from James McGlinchey of RA Enterprises and Ralph Stamerjohn: o You can set a byte in system common to disable context switching. "It's your machine in user mode." But be careful - no ACP functions! o Use ASTs to keep real-time response high. - 89 - RSX/VMS RELATED TOPICS Page 10-4 Mathematics of RSX11M o Use pool wisely. Run VMR from a command file, not interactively, else it will fragment pool. Pre-allocated I/O packets are not allocated, until they are used. ___ McGlinchley chooses a large number (60), and has [1,2]STARTUP.CMD run a task that does that many QIOs to NL: to allocate them. Ralph thinks 4 is a good number; the goal is hit ratio. o In drivers, don't IOFORK if you can help it. Try to fork only for IODONE. o To debug a resident library, patch in "breakpoints" (octal 777). o SETEF does not declare a significant event. Neither does a ___ clock interrupt. A MRKT$ expiration does. RMDEMO causes many SEs. o Do not use the terminal driver for a non-terminal device or performance will suffer. o Suggested task priorities - BRU, DSC - 40 General utilities - 50 PIP - 60 EDT - 70 UFD - 80 Task loader - 240 TKTN - 245 etc. The "holes" leave room for real-time applications tasks. - 90 - CHAPTER 11 SITE MANAGEMENT AND TRAINING SIG Panel The Site Management and Training SIG is now chaired by Doug Bickford, formerly of the Documentation Working Group. He intends to conduct a documentation survey, and also requested library submissions. Topics discussed included: o RM05 head crashes - one site's story. They completed mass replacement of RP06 drives with RM05s and have been sorry. Of 114 new RM05s, 20% have failed due to causes other than media. Rumor that DEC is prodding CDC. Make sure the legs ____ ____ ___ ____ are screwed down or the resulting vibration will torque the ___ _______ ____ bearings. They found that Field Service was improperly installing the drives. o DEC RM05 disks have sector slots on the bottom. There are two types - long slot and short slot. Some drives can't read one of the types, giving symptoms such as "Boot block not found", etc. o RM80 disk drives have excellent reliability even in a dirty environment, but clean the filters often. _____ ___ _______ _____ o Disk pack cleaning - recommend inspection 3 times per year, cleaning once per year. DEC Field Service will soon begin offering this service for 1-2 platter disks, and later will service larger packs. They will clean the packs every 6 months. o DEC has established a free "Upgrade Help Hotline" for customers only. The number is (800)343-4040. o Establish and adhere to a change control system (log) for your site. o For medium to large sites, a hierarchy of System Manager -> Operators -> Users has been found to work well. - 91 - SITE MANAGEMENT AND TRAINING Page 11-2 DEC Education & Training DEC Education & Training Digital Ed Services sponsored a workshop to discuss new products, and to advise on the best use of training dollars. In times of budget cutbacks, training frequently gets the axe. When making a case to management, remember that training is a service; one cannot measure productivity increase in dollars. Very few sites budget the training costs up front. After spending large sums on hardware and software to solve a particular problem, don't be reluctant to give the people what they need to ensure success. Try to view training as insurance - a protection against personal risk. ____ New and current products - o The Operator self-paced course is very good (slide show and workbook). The student must already know a little about computers. You must have a local course administrator for ____ this method to be effective. Also, do not expect students to complete this work on their time. Set specific milestones, including hands-on tasks. o VMS CAI course is available, including a VAX User Intro workbook. o New seminars on - VMS Application Optimization VMS Tuning DATATRIEVE generic (RSX or VMS) o A DATATRIEVE CAI course is in the works. o The new Courseware Authoring System (C.A.S.) was wrung out at 12+ test sites during 1982. C.A.S. will be integrated with video disks at a later date. Also planned are development aides, and consulting contracts. - 92 - CHAPTER 12 COMPUTER SECURITY Security Issues in VMS Access Control This session was presented by Gerry Smith of the VAX Development group (VSIG pp. 359-360). She outlined a "proposed" system for Access Control Lists (ACLs) as they "might be implemented in a future major release of VMS." As envisioned, the ACL would be an extension of the UIC file protection concept. An ACL is a list of identifiers and associated access rights. These rights include the current RWED plus a new OWNERSHIP which gives one the right to change the ACL. ACLs may be applied to various objects, including, but not limited to, files. The access rights list is associated with each agent (process) and would be managed via AUTHORIZE or via system services. An additional level of security will be implemented by allowing an object to be classed by security level and integrity level. Security level refers to the sensitivity of information and will range from 0 as the least sensitive to 255 as most sensitive. The object of security levels is to prevent disclosure of information; therefore, to READ an object, one would need a security level >= the objects ("read down"); to WRITE an object, one would need a security level =< the objects ("write up") to prevent a higher classified process from writing into a less-classified object. Conversely, integrity level refers to the trustworthiness of the information, with 0 being least trustworthy and 255 most trustworthy. The object of integrity levels is to control modification of data; hence, to READ an object requires integrity levels =< the object, while to write it requires integrity levels >= the object. For example, SYSTEM files would have high integrity, but low security. Every VMS routine which allocates memory (e.g. LIB$GETVM) zeros the memory space before giving it to the user. There will also be additional privilege bits implemented. In all cases, the ACL serves as an exception to the UIC protection. _________ That is, if UIC check passes, no further checks are made. - 93 - COMPUTER SECURITY Page 12-2 Security Issues in VMS Access Control In the Q&A session that followed, Gerry suggested that it might be possible to give access rights to a "trusted" program, analogous to INSTALLing a program with privilege. Global sections may also have security levels attached. Some type of logging of violations is in the works, but we were unable to get any details of the nature of this. For some time, users have requested the support of "execute-only" command procedures. This would provide the same protection currently possible with executable images. DEC is still stating that this is "very difficult", almost "unfeasible" within the VMS framework. Network Security - Proxy Logins This session was presented by Gerry Smith of the VAX Development Group. She discussed the concept of proxy logins, an unsupported, undocumented feature which is latent in VMS and DECnet-VAX. It appears there so that DEC personnel can experiment and refine the idea. (Since the implications to the DECnet product are far reaching, the support of this or any other method would require the approval of the Digital Network Architecture (DNA) Group.) Gerry describes proxy login as "a direction, not the solution" to enhanced network security. Why proxy? When one attempts a remote file access using DECnet-VAX, the remote node creates a process which provides a context for the File Access Listener (FAL). This context is established by a username/password combination which may be explicitly specified in the remote request, or defaulted to the NETNONPRIV account. In the latter case, the remote files cannot be accessed unless their protection grants the necessary privileges to the WORLD. In the former case, the requester must send a password in the clear (unencrypted). More so, the password must be typed in the open, or even appear in a command procedure. When proxy logins are enabled on the source and sink nodes, the source may make the remote request using a new format of access control string (ACS). When the requester specifies an ACS, the old rules apply. However, when the requester omits the ACS from the node spec, DECnet-VAX sends only the username to the remote node. The LOGINOUT image on the remote node recognizes that proxy login is desired, and instead of using SYSUAF.DAT - 94 - COMPUTER SECURITY Page 12-3 Network Security - Proxy Logins for validation, it uses NETUAF.DAT. This ISAM file contains records of the form: :: If the specified record is found, the process is created in the context of the local username, including UIC, privileges, and other attributes specified in SYSUAF.DAT. It is important to note that NETUAF provides a 1:1 or N:1 mapping, but not a 1:N mapping (duplicate keys are not allowed). The remote user may force the use of the default NETNONPRIV account by specifying a null ACS (""). Limitations The usual caveats apply. If a user has the ability to change his username (CMKRNL privilege), he may impersonate someone else both locally AND remotely! The proxy scheme lacks authentication and depends heavily on individual node security. One must trade these shortcomings against the significantly higher possibility of compromise with the present system. Turning it on Four VMS images have been modified to implement proxy logins -- NETACP, NCP, LOGINOUT, and AUTHORIZE. o LOGINOUT - described above o AUTHORIZE - - A new command has been added to AUTHORIZE: CREATE/PROXY creates [SYSEXE]NETUAF.DAT - The SHOW, ADD, and REMOVE commands now accept the "/PROXY" switch To see the syntax of these new commands, use the LIBRARIAN to extract all modules from the SYS$HELP:UAFHELP.HLB library: $ LIBRARIAN /EXTRACT=* SYS$HELP:UAFHELP /HELP Then edit UAFHELP.HLP and look for the sections that are commented out. Uncomment them, and replace the library. o NETACP - supports the proxy access attribute for both nodes and objects, and recognizes connect requests of the proxy format (type 2). - 95 - COMPUTER SECURITY Page 12-4 Network Security - Proxy Logins o NCP - supports new command qualifiers: - SET EXECUTOR PROXY ACCESS {none|incoming|outgoing|both} - SET NODE n PROXY ACCESS " " " " - SET EXECUTOR DEFAULT PROXY ACCESS " " " - SET KNOWN OBJECTS PROXY ACCESS " " " If you really want to know the details of proxy logins, spend some time wading in the "fiche" pond. NOTE Remember, this is an undocumented, unsupported feature! If you decide to experiment, you do so at your own risk. Don't call DEC. But if you would like to share your assessment of this new idea, drop a note to the newsletter editor of the VMS Security Working Group: C. Douglas Brown Distributed Processing System Design Division 2644 Sandia Laboratories Albuquerque, N.M. 87115 VMS Security Cal Page, a VAX computer security expert, gave an enlightening talk on VMS security. While in the exhibit area, Cal performed his own security audit of one VAX system. On a scale of 0 to 10, he gave it a 2. The high points his presentation follow: Common VMS attacks and counters o Guessing passwords: by intelligent guessing, an attacker ________ _________ attempts to login. Counter: Instruct users to select obscure passwords with _______ at least 6 characters (get the patch on the Fall SIG tape). Change passwords periodically, especially for privileged accounts. Restrict dialup access to certain accounts, and restrict hours for accounts (e.g., no weekends). Use a terminal snatcher program to intercept logins, and do additional authentication, auditing, or journaling. DEC - 96 - COMPUTER SECURITY Page 12-5 VMS Security Software Services will give you a patch to enforce a login failure limit. o Trojan horse: a patch, typically in a game, that attempts ______ _____ to store a record into SYSUAF. If that program is run by someone with sufficient privilege, the patch succeeds. Counter: check protection on files often; also scan _______ SYSUAF. Compare CDD ACLs to the file protection on the databases. Compute checksums for utilities and compare them periodically. Encrypt the checksums if necessary. If you disallow games on your system, people will likely revive them under a disguised name. Instead, make a separate games account, with NO privileges, and disallow file access to other than OWNER. You could also restrict CPU usage, and time of day, etc. o Login fake: You are running some program, and it _____ ____ mysteriously aborts and appears to kill your process. It then prompts you for "Username:" and "Password:" and you happlily comply. Alternate scheme - previous person leaves program running, faking a logout display on the screen. Once again, you mistakenly "login". Counter: Instruct the users to notify the system manager _______ of any abnormalities in the login procedure. Use a terminal watcher program, which will automatically logoff inactive processes. General tips o Field Service: protect SDA from F/S, since it can be used _____ _______ to search typeahead buffers for passwords (^Y flushes the buffer). If you use the RDC, hide your RDC modem phone number. Try dialing your modem; if you get an answer tone, it's bad news. Remember that through the console, one can set bits that filter out any RDC traffic from printing. o Guard the OPA0: console for the above reason. Also, one could ^P the machine, and do a conversational boot to prevent VMS from using the correct UAF. Do not separate the console printout; scan through it periodically. If necessary, leave the machine keyswitch in the console disable position. o Use a scanner program to locate processes with elevated privileges, and generate an audit trail. - 97 - COMPUTER SECURITY Page 12-6 VMS Security o Check logical names and make sure they point to the right places. o Make sure that priviliged images are LINKed with /NOTRACE/NODEBUG. If you do not have the program objects, you must write a program to "zap" the image file traceback vector. Note that this is currently a problem with several popular canned packages such as PATRAN and EUNICE. o Grant privileges with the utmost care. Cal thinks that "privilege" was a poor name choice, since it is status oriented. Note that KE780 microcode development requires CMKRNL. o Do not use terminals with "soft" keyboards which permit function keys to be programmed to generate arbitrary character sequences. Someone may send an innocent-looking MAIL message with codes embedded to program a key and execute it. Lost data issues Cal's advice on data backup and storage: o To pick a backup interval, ask yourself what you can afford to lose. o Use more than one set of tapes, and rotate them. o Backup using the grandfather/father method - Do a full backup, followed by some number of incrementals, and repeat. Store the full backup some place else. (We backup once a week, with full backups once a month. Each incremental is started on a new tape, for ease of recovery. Tape sets are saved for one year, then reused.) o Run with the backup copy so that you know it's good. This also accomplishes disk compression. o Always backup immediately prior to a disk or tape PM. Never let someone perform tape and disk alignment at the same time! - 98 - COMPUTER SECURITY Page 12-7 Miscellaneous Tidbits on VMS Security Miscellaneous Tidbits on VMS Security The following pieces of information fit into no specific structure and were gleaned from private conversations with VMS developers and other members of the Security Working Group. o There already exists an "erase-on-delete" bit in the file header. This will cause a file's blocks to be set to zero when the file is deleted. NOTE While bearing in mind that this is an unsupported feature, we couldn't resist trying out some of what we heard. Thus, if one types "SET FILE/ERASE filename" and performs a "DUMP/HEADER filename", one will find that DUMP knows how to interpret the bit. Performing a simple DELETE on this file does NOT however, appear to zero the file. If one performs a "DELETE/ERASE filename", the file is overwritten with zeros prior to deletion. This occurs regardless of the state of the erase-on-delete attribute bit in the file header. This was verified by creating a contiguous file, noting its retrieval pointers, performing a DELETE/ERASE, mounting the volume FOREIGN, then DUMPing the LBNs previously given. The blocks contained all 0's where they previously had other information. o In one of the security sessions, mention was made of an (undocumented) flag in the FAB which disables logical name translation for that file (actually, it forces use of only the SYSTEM logical name table). This would ensure that someone executed the program intended, and could not redefine a logical name to execute something else. In a later session (VMS Advanced Q&A) Tim Halvorsen & Dick Hustvedt warned that this will probably not work in future _ releases! o The documentation set for the "next major release of VMS" will have a separate "security manual" which will detail all of the options for system managers. It should also detail what files need what kind of protection and provide other information needed by system managers to make system penetration more difficult. - 99 - COMPUTER SECURITY Page 12-8 Miscellaneous Tidbits on VMS Security o There will be a patch to the SET PASSWORD routine (on the Fall SIG tape) which will ensure that a password contains a specified minimum number of characters. - 100 - APPENDIX A VAX/VMS SIG FALL 1982 HANDOUTS TABLE OF CONTENTS VAX Systems SIG Roadmap 1 VAX-11 Pascal Features 7 V2 BASIC 14 VAX-11 COBOL Features Overview 34 * COBOL Report Writer Facility 42 * Structured Programming in COBOL 58 VAX-11 RMS Tutorial 64 * Performance Management Techniques 67 + Semi-auto Conversion PDP to VAX Assembly Language 74 VMS Performance Tools 84 + Exec Enhancements to VMS V3 87 * + Real-Time Processing for New VAX User 94 * Coding Changes for V3 108 * Network Security 114 + Using VAX BASIC to Develop for PDPs 115 VAX Cluster Concept 117 * + Real-Time Aircraft Simulation 119 + Data Acquisition System for Digital Image Proc 140 Interactive Digital Signal Processing 142 Justifying a New VAX 154 VAX-11 Condition Handling Facility 161 VMS Lock Manager 166 * COBOL Optimizations 173 VAX-11 PL/1 Futures 188 Record Structures in BASIC V2 193 VAX DEC/CMS 199 VAX-11 DEC/CMS:A Tutorial 205 Using the Common Data Dictionary from BASIC 208 Using USEROPEN in V2 BASIC 218 A Tutorial on MDL Structure Definition Translator 227 + Modular Programming on VAX/VMS 232 Magtape Tutorial 240 * + INDAS: Integrated Data Analysis 248 _______________ + Sessions attended * Duplicate handouts from Spring 1982 DECUS - 101 - VAX/VMS SIG FALL 1982 HANDOUTS TABLE OF CONTENTS Page A-2 VMS Backup 285 * + What's in V2 BASIC Kit? 287 Intro to VAX/VMS System Services 296 VAX/VMS on 11/730 310 * + Multi-Lingual Commercial Applications with COBOL 314 Intro to Run Time Library 325 * Replacing Object Libraries with Shareable Images 333 + VMS Terminal Driver Performance Improvement 339 * Conversational SAAM: A Modeling Language 346 + Monitoring VAX/VMS Security 348 + Data Compression Disk Driver 349 Data Encryption Disk Driver 351 VAX-11 RMS Utilities Tutorial 353 * + Security Issues in VMS Access Control 359 + VAX DECNET Internals Tutorial 361 * VAX Performance Management 375 + Digital I/O via LPA-11K 378 New Mass Storage Architecture 381 + Shareable Images on VMS 392 * VMS Utilities 399 * We Do It with Mirrors 415 Introduction to DCL 417 Maintaining our Systems 420 Introduction to VAX/VMS System Management 423 _______________ + Sessions attended * Duplicate handouts from Spring 1982 DECUS - 102 - APPENDIX B NETWORKS SIG FALL 1982 HANDOUTS TABLE OF CONTENTS ETHERNET - A Technical Overview 1 (Development of a Local Area Network) ETHERNET (CSMA/CD) Performance 25 + (Performance of a Simulated Programming Environment) Technologies for Local Area Networks 37 (Local Area Network Technology Comparison) Operational Experiences at Brown University (BRUNET) 79 Using Broadband with Digital's Network Products 60 (Broadband Communications Capabilities) Network Communications Hardware 86 DECnet/SNA Gateway Overview 102 + Programmer's View of RSX DECnet 110 DECnet-VAX Internals 146 * Digital and Packet Switched Networks 168 Monitoring Network Status and performance 186 + Introduction to Networking 212 + (Distributed Processing and Networks) DECnet Introductory Tutorial 226 + Networks and Communications Update 240 + _______________ + Sessions attended * Duplicate handouts from Spring 1982 DECUS - 103 - APPENDIX C DATA MANAGEMENT SIG FALL 1982 HANDOUTS TABLE OF CONTENTS The Process of DBMS Evaluation 1 Virtual Record Locking in RMS under RSX-11M/M+ 5 Real-time Automated Inventory Control with RMS 10 VAX-11 DBMS Technical Tutorial 18 VAX-11 Common Data Dictionary 41 A Comparison of INDENT and FMS 52 Data Management Tools in Scientific & Engineering 60 Environments Data Base Modeling and Analysis Techniques 64 VAX-11 DBMS Data Base Design and Tuning 75 Data Base Security 85 - 104 - Page Index-1 INDEX ACCOUNTING DEUNA, 7-2 Image-level, 2-3 Performance, 7-2 ACL, 12-1 EUNICE, 12-6 ACP, 2-5 ANALYZE/RMS FAB, 12-8 bugs, 2-8 FDT, 8-6 APL, 4-2, 5-1 Field Service, 12-7 AUTOGEN, 2-6 FMS, 5-9, 8-6 FORTRAN, 5-1 BACKUP, 2-5, 3-1, 4-2, 4-8, 9-9 Furniture/PC, 8-10 Backup policy, 12-7 BASIC 2.0, 4-2 Gateway, 7-3 3270, 7-4 C, 6-1 RJE, 7-4 CI780, 4-2 GKS, 9-3 Cluster, 9-10 CMS, 6-1, 6-3 HSC50, 9-7 Colorado Support Center, 7-8 Computer-Aided Instruction, 11-2 IBM, 7-3, 8-9 CONVERT, 2-8 3270, 7-4 Courseware Authoring System, IDC, 2-9 11-2 Initialize tape, 4-8 CP/M, 8-9 Integrated Disk Controller, 2-9 DATATRIEVE, 5-3, 7-6 KE780, 12-6 CROSS, 5-4 to 5-5, 5-7 Key compression, 2-9 Date arithmetic, 5-6 FIND, 5-3 LA100, 8-2 to 8-3, 9-5 DEC/CMS, 6-3 LA50, 8-2 to 8-3 DEC/MMS, 6-1 LIB$SIG_TO_RET, 4-9 DECnet, 7-1, 8-8 LN01, 4-2, 9-3 to 9-4 Monitor, 7-5 LOGO, 8-9 OBSERVER, 7-5 LP27, 4-2, 9-5 RSX 3.1, 7-8 LQP02, 8-2 TSC, 7-8 DELETE/ERASE, 12-7 Mass Storage Architecture, 9-10 DEUNA, 7-2 MIL-SPEC, 9-1 Disk pack cleaning, 11-1 MMS, 6-1 DMF32, 3-2, 4-2 MONITOR, 2-2 DMR11/DMC11, 7-8 MSCP, 9-10 MUMPS, 8-9 ECO 6, 4-8 MWAIT, 4-8 ECO 7, 2-7, 4-8 EDT, 7-8 Network Security, 12-2 Erase-on-delete, 12-7 Norden, 9-1 Ethernet, 7-2, 8-8 - 105 - Page Index-2 OBSERVER, 7-5 SET FILE/ERASE, 12-7 SET PASSWORD, 12-8 P/OS, 8-5 SHUTDOWN, 3-2 PASCAL, 4-2, 5-3 SIGGRAPH, 8-3, 9-3 Password, 12-5 SNA, 7-3 PATRAN, 12-6 SYSAP, 9-10 PDP-11, 7-4 SYSGEN Parameters PFCDEFAULT, 2-4 FREELIM, 2-6 Preventive maintenance, 12-7 PFRATL, 2-6 PRO-300 Personal Computers, 8-1 POOLPAGING, 2-6 FDT, 8-6 SWAPOUTPGCNT, 2-6 FMS, 8-6 SYSMWCNT, 2-6 Graphics, 8-2 SYSPAGING, 2-6 P/OS, 8-5 TBSKIPWSL, 2-7 PRO-Basic, 8-6 to 8-7 TTY_SILOTIME, 3-2 SORT, 8-6 UDABURSTRATE, 9-6 Telephone Management System, 8-4 TA78, 9-8 Tool Kit, 8-6 Telephone Management System, 8-2 UCSD P-System, 8-7 TEMPEST, 9-1 Prologue 3, 2-9 TMS, 8-2, 8-4 Proxy login, 12-2 TU78, 9-8 TU80, 4-2, 9-9 QUALIFY, 2-11 UBA, 9-12 RA60, 4-2, 9-8 UDA50, 9-5 RA80, 9-8 UNIBUS RA81, 9-7 timing, 9-11 RDC, 12-6 UNIBUS Adapter, 9-12 ReGIS, 8-4 Buffered data path, 9-12 RFI/EMI, 9-1 LWAE, 9-12 RJE, 7-4 UNIX, 6-1 RM05 MAKE, 6-1 head crash, 11-1 PDP-11, 6-1 RM80, 11-1 SCCS, 6-1 RMS32, 2-8 Upgrade Hotline, 11-1 bug, 2-8 RSX11M+, 8-5 VAX 11/730, 2-9 VAX 11/782, 2-11 SCA, 9-10 VAXStation, 9-2 SCS, 9-10 VMS 3.1, 4-2 SDA, 2-3, 12-6 VMS 3.2, 3-1 to 3-2, 4-2, 9-9 SEARCH, 4-8 VMSINSTALL, 4-2 Security, 12-1 VNX, 6-1 Access Control List, 12-1 VOLSET Field Service, 12-6 corrupt, 4-9 Network, 12-2 VS100, 9-2 Proxy Login, 12-2 VT100, 9-1 RDC, 12-6 Bugs, 9-11 SDA, 12-6 VT125, 8-2, 8-4, 9-1, 9-3 - 106 - PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX Systems SIG Spring 1983 SIR List VAX Systems SIG Spring 1983 SIR List Attached you will find a collection of System Improvement Requests and a ballot form on which to record you preferences. The SIR's are a major vehicle by which the VAX user community expresses its concerns and desires to Digital. Your participation in this process is most important. Please take the time to review the enclosed SIR's and assess their effect on your use of VAX's. We then ask you to rank your favorites by assigning a point value to them. Please also fill out the enclosed questionaire to allow us to establish demographic patterns to the responses. Please return your ballot AS SOON AS POSSIBLE so that the results may be tallied and presented to DEC. The results of ballotting and some form of response by DEC will be given at the VAX SYSTEM SIG SYSTEM IMPROVEMENT REQUESTS session of the Spring DECUS Symposium. Instructions You have a total of 100 points to allocate among the SIR's on the ballot. The more points you allocate to a particular SIR, the more strongly you wish to see this feature included in the VAX or VMS. You may assign your points in either a positive sense (to encourage the change) or a negative sense (to discourage the change). In order to assure a wide range of improvements, we have limited the number of points that may be allocated to any SIR to 10. To allocate points to a SIR simply record the number of the SIR in the column labeled SIR NUMBER, and the number of points to allocate to it in the column labeled POINTS. Remember only 100 points total (absolute value). Please note that any ballot not following the points assignment rules, or not specifying a DECUS membership number will not be counted. Only one ballot per member will be accepted. - 107 - PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX Systems SIG Spring 1983 SIR List SIR: S83-001 Abstract: Provide a terminal driver I/O request to insert characters into the beginning of the typeahead buffer. Description: This I/O request would be extremely useful in applications which mix RMS and QIO handling of terminal input. For example, programs which interrogate a terminal to determine its attributes could first read any typeahead, interrogate the terminal, and then restore the previous typeahead contents. This approach would be much more modular than any current alternative. SIR: S83-002 Abstract: Provide more complete support for operator handling of mountable devices. Description: It should be possible to operate VMS in a mode which excludes user requests of specific disk or tape devices. All requests for device assignment should be considered generic. The current capabilities of operator-assisted mount are not sufficient since it does not extend operator control over the INITIALIZE and ALLOCATE functions. Also, a site specific "slot number" or "visual id" field needs to be available with all volume requests and must be propagated to ALL operator messages relating to the volume. SIR: S83-003 Abstract: The name of the queue in which a batch job is executing should be available via $GETJPI. Description: Many sites use costing schemes with rates which differ for different queues. There needs to be an easy way to obtain the identity of the queue from which a job is executing, while the job is executing. SIR: S83-004 Abstract: Multiple levels of control are needed for terminal BROADCAST messages. Description: There need to be at least three levels of control over BROADCAST messages to terminals. These levels are: never send a BROADCAST message (this is not an interactive terminal), send only "high priority" messages (allows system manager to warn that system is going down, etc.), allow all BROADCASTS (MAIL, /NOTIFY, etc.) - 108 - PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX Systems SIG Spring 1983 SIR List SIR: S83-005 Abstract: Provide a process specific time-of-day clock. Description: There should be a process unique time-of-day clock setable (with privilege) to a date and time of the user's choice. This would allow for testing of applications algorithms which are time of day dependent independent of the system time. It would also allow a user remote from the system to deal with system time relative to his local time zone. Finally, it would allow process specific rollback of time to duplicate files or database data generated in the past. SIR: S83-006 Abstract: Provide for control of VMS scheduling behavior on an image or process basis. Description: It would be useful for certain applications to be able to control scheduling parameters such as priority boost for a process or image. For example a process which was able to sustain high CPU utilization while performing direct I/O's might be denied the direct I/O priority boost to prevent it from gaining an unfair portion of the CPU. SIR: S83-007 Abstract: Provide a means of disabling automatic expansion of the user-mode stack. Description: When developing real time applications, and locking the user-mode stack in memory, it would be nice to know that the stack size specified was sufficient and that additional unlocked pages where not being allocated. SIR: S83-008 Abstract: The $GETJPI system service should return a process default device and directory. Description: These two pieces of information are very useful. Particularly, this would allow locating a user's default directory from a Process ID. SIR: S83-009 Abstract: Image accounting records contain untranslated concealed device names. - 109 - PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX Systems SIG Spring 1983 SIR List Description: Since the values of concealed device names may change from time to time, it is problematical if SYSTEM hidden device names are better off translated or not. For non-SYSTEM logical names, translation should normally occur. For example, if a GROUP or USER concealed logical name duplicated a system logical name, it would be impossible to tell the actual location of the image represented by the accounting record. By default, SYSTEM names should not be translated; USER and GROUP names should. However, there should be separate switches for each mode. SIR: S83-010 Abstract: Improved documentation of VMS system service capabilities. Description: There should be an Applications manual as part of the VMS documentation set. The manual would present topics according to application or use, as opposed to system service. It should also present topics in a "How to" format and should be oriented toward high level language use of the system services. SIR: S83-011 Abstract: Improve handling of CPU quota exceeded error. Description: When the CPU quota is exceeded, the current image is terminated, a message is displayed, and the user is returned to DCL with an extended time limit. The user should be able to resume executing the image by entering $CONTINUE (most desirable) or should receive a warning prior to the expiration of the COU limit. Also, the quota expiration message should always appear. Currently, if the quota is exceeded while executing in DCL, no message appears. SIR: S83-012 Abstract: An additional CPU time limit is needed for images. Description: In addition to the current CPU quota, there needs to be a time limit associated with the execution of images. There is a distinction between the amount of CPU time authorized for a session and the largest amount needed to run an image. This limit should be associated with a process and be setable from DCL. Such a feature would be particularly useful in an educational environment where it could be set low for the purpose of trapping runaway student programs. This limit should be implemented as a - 110 - PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX Systems SIG Spring 1983 SIR List condition so that it can be processed via a handler and so that it is useable by such images as the BASIC interpreter, where there can be multiple program "runs" for one image activation. SIR: S83-013 Abstract: The DCL SPAWN command needs /CONTINUE qualifier. Description: One of the most common uses of the SPAWN command is following the Control-Y interruption of an image. It should be possible to perform the SPAWN with a /CONTINUE which indicates that the interrupted image is to be CONTINU'ed following a return from the subprocess. This would prevent accidental loss of the executing image (which was probably a text editor). SIR: S83-014 Abstract: Extend the DSBMSK parameter to the F$LOGICAL() lexical function. Description: The DCL lexical function F$LOGICAL should accept an additional argument restricting the logical name tables to be used for translation, just as the $TRNLOG system service does. This would allow canned DCL procedures to be sure that they are using the appropriate system wide values for logical names. In addition, ths would allow DCL procedures to be able to readily restore logical names to their original values. Currently, a DCL procedure can attempt to save the original value, but DCL has no way of determining if the equivalence name is a PROCESS, SYSTEM, or GROUP value. SIR: S83-015 Abstract: Improve support for foreign commands. Description: The contents of a foreign command line should be available without the case conversion normally done by DCL. Also, the limit on the length of a foreign command line is overly restrictive. The maximum length should be setable like the DCL symbol table size. SIR: S83-016 Abstract: Enhance the INQUIRE command to reduce the need for editting of responses. Description: The INQUIRE command should be enhanced to have at - 111 - PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX Systems SIG Spring 1983 SIR List least the capability of the MCR .ASK, .ASKN, and .ASKS. This makes the migration from MCR to DCL much easier. It also saves considerable coding in DCL. INQUIRE should provide for checking of responses for such things as YES/NO, string, and numeric (with range and radix checking). If the response does not meet the checks, the question is automatically repeated. It should also be possible to specify default values to be taken, and to support timeout of the INQUIRE. SIR: S83-017 Abstract: Provide a network COPY symbiont to allow store-and-forward transmission of files to any network node. Description: The symbiont would transmit the file when the appropriate network link became available. The mechanism should support a /AFTER qualifier to allow taking advantage of differential time-of-day data transmission rates. Such a mechanism would be much cleaner than any present alternatives. SIR: S83-018 Abstract: The VMS COPY command needs a /OWNER=uic qualifier. Description: Frequently, files are created by one set of users on a system for use by another set. These files must be "delivered" to their end users with the correct ownership. The COPY command should have a /OWNER qualifier which allows a suitably privileged user to specify the ownership of an output file. SIR: S83-019 Abstract: A tape transfer utility is needed. Description: A tape handling utility should exist in VMS to read and write non-VMS tape formats. This is especially useful when conducting projects which encompass the use of more than one computer. Presently available utilities will not allow transfer between DEC supported operating systems, e.g. VMS, RT11, and TOPS-10. SIR: S83-020 Abstract: Implement the capability to do a DIRECTORY by UIC. Description: It is frequently necessary to locate all the files - 112 - PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX Systems SIG Spring 1983 SIR List on a disk owned by a particular user. This is especially important in resolving problems involving disk quotas. The DIRECTORY command should have a qualifier which would allow a selective listing of files by owner UIC. SIR: S83-021 Abstract: Compilers and utilities should be able to reuse existing output and listing files. Description: A performance gain could be obtained if the VMS compilers and utilities could be persuaded to reuse existing existing output, listing, and map files when explicit version numbers are specified. The current behavior is to delete the file and recreate it. SIR: S83-022 Abstract: The VAX FORTRAN compiler should provide controls for listing formatting. Description: The VAX FORTRAN compiler should provide a means for controlling page ejects in source listings. The ability to provide titles would also be useful. SIR: S83-023 Abstract: The VMS RTL should provide general purpose menu processing. Description: Menus provide a very simple means for controlling production and application software. The VMS Run Time Library should provide a fairly high level menu processor. Menus can be done from FMS, but too much work needs to be repeated for each application. SIR: S83-024 Abstract: Disk bad block reporting should be more explicit. Description: If the error logger detects a bad block in an existing file, it will continue to report errors for each access to the block. Only if the file is deleted will it be flagged as bad and put out of service. While the system must know where the bad block is, it does not provide enough information to determine the file in which it is contained. Only with this information is it possible to do something about the recurring errors. - 113 - PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX Systems SIG Spring 1983 SIR List SIR: S83-025 Abstract: Automatic file encryption. Description: Provide optional, automatic file encryption where the encryption key is provided when a file is opened. This assures that uncoded data is never tranferred to disk (or backup tape) where unauthorized access is more likely. It would provide a common encryption mechanism that was readily available to all applications. SIR: S83-026 Abstract: Provide recognition style input for file specifications. Description: The TOPS-20 operating system provides a mechanism for specifying filespecs in which only unique stems of each filespec component need be entered. A control character is entered following the stem, signalling the operating system to recognize and complete that part of the specification. This feature would be useful in VMS when long filespecs are used. SIR: S83-027 Abstract: Provide some form of file compression. Description: The VMS file system should provide an optional mechanism for compressing files for which compression is a benefit (such as line printer listings). Access to the compression should also be provided as a qualifier on the COPY command. File compression, used when appropriate, can reduce disk space usage and time required for transmission of files over DECnet. SIR: S83-028 Abstract: Remove the name length restriction on wildcard file specifications. Description: Wildcard searches specified using more than the normal limit of characters for RMS file names are not supported. Currently, if more than 9 characters are specified in a filename or more than 3 characters in a filetype, RMS indicates an error. For example, the filespec *DATABASE*.DAT should be allowed since it can match legitimate file specs (the *'s can match zero length substrings). - 114 - PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX Systems SIG Spring 1983 SIR List SIR: S83-029 Abstract: There should be full support for print and batch jobs in VAX to VAX networks. Description: There should be full support for directing a file to a print or batch queue which resides on another VAX node in a DECnet network. All options that can be specified for a local job should be supported. This simplifies having certain machines in a network dedicated to batch or applications processing. The work can be conveniently routed to the appropriate batch queue. This also simplifies routing of printed reports, etc. to various destination nodes. - 115 - PAGESWAPPER - February 1983 - Volume 4 Number 4 This page left intentionally blank - 116 - PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX Systems SIG Spring 1983 SIR Ballot VAX Systems SIG Spring 1983 SIR Ballot Questionaire DECUS membership no. ____________ Installation Delegate? YES NO Your experience level (circle 1): 1. Wizard 2. Expert 3. Knowlegable 4. Normal 5. Novice -------------------------------------------------------- Configuration _____________ CPU Type (circle one; if more than one, indicate number of each): 1. 11/782 2. 11/780 3. 11/750 _____ _____ _____ 4. 11/730 _____ For Primary System: Mem. Size (Mb): System Disk: _____________ ________________ Tot. Disk Space (Mb): No. of Users: _______ _______________ VMS Version No.: Avg. User Load: ____________ _____________ Operating System (if not VMS) or ________________________ Operating System Migrating From -------------------------------------------------------- Application Area (circle all that apply) ___________ ____ 1. End User 2. H'ware OEM 3. S'ware OEM 4. Sys Integrator 5. Enginerring 6. Research 7. S'ware house 8. Service Bureau 9. Scientific 10. Business EDP 11. Government 12. Utility Co. 13. Turnkey Sys. 14. Manufacturing 15. Education Principle Language ____________________ Product Line (if known) _______________ - 117 - PAGESWAPPER - February 1983 - Volume 4 Number 4 VAX Systems SIG Spring 1983 SIR Ballot VAX SYSTEMS SIG ___ _______ ___ SPRING 1983 SIR BALLOT ______ ____ ___ ______ Tally Reminder: The total number of points (absolute value) which you may allocate to all of the SIR's must be no greater than 100 points, and no more than 10 points may be allocated to any single SIR. SIR Number: Points: ___ _______ _______ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ ____________________ Mail to: Mr. Gary Grebus Battelle Columbus Laboratories Room 116011 505 King Avenue Columbus, Ohio 43201 - 118 - PAGESWAPPER - February 1983 - Volume 4 Number 4 INPUT/OUTPUT Submission Form A SIG Information Interchange Please reprint in the next issue of the Pageswapper Caption: ______________________________________________________ Message: ______________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ Contact: Name _______________________________________________________ Address ____________________________________________________ ____________________________________________________________ Telephone ____________________________ If this is a reply to a previous I/O, which number? ________ Signature _____________________________ Date ________________ Mail this form to: PAGESWAPPER Editor, DECUS, MR2-3/E55, One Iron Way, Marlborough, MA 01752, USA - 119 - PAGESWAPPER - February 1983 - Volume 4 Number 4 Tear out to submit an I/O item PAGESWAPPER Editor DECUS, MR2-3/E55 One Iron Way Marlborough, MA 01752 USA - 120 - PAGESWAPPER - February 1983 - Volume 4 Number 4 System Improvement Request Submission Form SIG ref no. _________ Page 1 of _____ ________________________________________________________________ Submittor: Firm: Address: Phone: ________________________________________________________________ Circle application area(s) most closely related to yours (OEMs circle end use): Transaction Processing Business EDP (accounting) Program Development Systems Development General Timesharing Student Timesharing Shared Small Applications Shared Large Applications Process Control Word Processing Large Simulation ________________________________________________________________ System Configuration: CPU Model: System Disk: Memory Size: Average User Load: Operating System: Version: ________________________________________________________________ Abstract (Please limit to four lines): ________________________________________________________________ Description (include justification and expected usefulness): Use additional pages if required Completed SIR should be returned to: Gary L. Grebus, Battelle Columbus Laboratories, 505 King Avenue, Columbus, Ohio 43201, USA PAGESWAPPER - February 1983 - Volume 4 Number 4 Tear out to submit an SIR Gary L. Grebus Battelle Columbus Laboratories 505 King Avenue Columbus, Ohio 43201 USA