An Interview With Mark Gorham Of OpenVMS

Ken Farmer writes "There's already been one press interview with Mark Gorham, but that encounter with HP's VP of the OpenVMS Systems Division omitted some technical details that warrant further attention. Hence, SKHPC thought it appropriate to go on a deep dive with one experienced in OpenVMS and SCUBA diving as well."

Comments

the reports of my death ... greatly exaggerated...

[lophophore]

VMS keeps coming back, and appearing on Slashdot like a bad penny. The IA64 has breathed new life into this OS, which is the most secure and stable that I have had the pleasure to use. VMS had a C2 security rating out of the box in 1990 or so, but what this article does not mention is that a variant version (SEVMS) carried a full B2 rating., which is really something.

Mark who? I don't know his name. I worked for DEC VMS Engineering in the VAX and Alpha days, who is this guy?

This article makes it seem like the idea of building unix apps on VMS is a new thing. It's not. VMS Posix was available in 1992, and many Unix/C apps would just compile and run. It was very cool.

The dinosaur is aging very well.

Re:New VMS users?

[VAXcat]

Reliability, scalability, uptime, high performance wide area clustering, no viruses, very few security problems of any kind (and those occur mostly in code migrated from unixland). A few of the reasons people choose VMS for an operating system. Individual VMS systems often have multi year uptimes (even in heavily used environments). VMS clusters have uptimes even longer still. And that's leaving out any of the religious flavored arguments about what OS is easier to administer and use.

Re:the reports of my death ... greatly exaggerated

[tshannon]

OK, a few facts for the record. I wrote the the Q's that Mark Gorham provided the A's to in the interview posted on my Web site. So I'm the perpetrator of what has turned into a Fine Mess.

I know plenty about SEVMS and its B2 security level rating as well as the circa-1992 VIP (VMS Integrated POSIX). I left this information out of the article because many of its intended readers don't know C2 from B2, and that VIP didn't cut it as a UNIX development environment. Better to keep things simple, the interview was long enough as it was. I didn't have the time to go down so many ratholes that an article became a book. (Been there, done that, didn't want to do it again yesterday.)

If VMS is a dinosaur, what's UNIX? It's an OS created 10 years prior to VMS, making it a Older Dinosaur. Neither of these dinosaurs are extinct, both have evolved. VMS can do things today that I had no clue it would be able to do today. Same goes for UNIX.

I don't know Mark Gorham's position or job title in the VAX and Alpha days, but he's currently the VP of HP's OpenVMS Division.

Cheers,

Terry Shannon

Re:How much of Dave Cutler's OpenVMS is left?

[tshannon]

The system implementation language for VMS was BLISS. The MACRO-32 assembly language was based thereon. The VAX hardware architecture and the VMS OS were co-developed and joined at the hip. As VMS evolved, portions of the OS were rewritten in C. When Alpha came along, the microcode that tied VAX to VMS was obsolete... VMS knew nothing of the Alpha architecture. Hence a hardware abstraction layer was used as a go-between. That was PALcode, or Privileged Architecture Library code. PALcode enabled VMS developers to rewrite MACRO routines in C. Not exactly the easiest of architectural ports, and it consumed the resources of the majority of the VMS Development team. Alpha to Itanium wasn't an easy port, but it was accomplished by about 25 developers over the course of 43 months. Far fewer lines of code had to be changed, and there was no code freeze... VAX/VMS V5.4 was the code freeze point for Alpha/VMS; while the Alpha porting team turned VAX/VMS V5.4 into the first Alpha/VMS release, the VAX/VMS developers kept on developing VAX/VMS. So VAX/VMS had more functionality (absent the 64-bit support) than did the initial Alpha/VMS release. It took a number of releases to bring Alpha/VMS up to feature 'n function level of VAX/VMS.

Haven't a clue if VMS was designed with clustering in mind right from the get-go, but VMS started shipping in ~1988, VAXcluster software materialized around 1983-84. Clustering definitely preceded VMS V5.0, the 5.0 release was all about a modular kernel and SMP support and security enhancements.

Dave Cutler did not write or design VMS, he was responsible for VAXeln, a run-time version of VMS. He then went on to develop MICA, the OS intended to run on the PRISM hardware architecture. PRISM was killed because the hardware existed, MICA was nowhere near ready for prime time, or even initial boot time.

MICA was designed to be a superset of VMS that reduced VMS limitations and expanded its capabilities. Cutler went to Microsoft the day after the PRISM project was cancelled, and he took the MICA code along with him. Cutler went on to develop NT, and DEC discovered that portions of NT were identical to portions of MICA, right down to the comment lines.

I'm not a lawyer, but I know enough about intellectual property to realize that NT contained a lot of DEC IP which DEC did not legally convey to Microsoft. DEC's IP lawyers knew that DEC had been ripped off. DEC knew that filing a theft of intellectual property lawsuit against Microsoft would be an exercise in futility, as DEC had far fewer lawyers and far fewer financial resources than did Microsoft.

The end result: the so-called "Alliance for Enterprise Computing." Big win for Microsoft, massive blunder on DEC's part. DEC's CTO, whose initials were BS, jumped at the first offer Microsoft put on the table. Bill Strecker knew a lot about computer technology and packaging, but he didn't have much in the way of negotiating skills. He jumped at an offer that sealed the fate of Alpha back in ~1994 or so.

Under the terms of the "deal," Microsoft agreed to endow Alpha with Intel parity on the server side, but not on the desktop. VMS minus desktop productivity tools (trivial things like the MS-Office suite) couldn't compete in the high-volume Wintel space. DEC unilaterally ceded a vast addressable market for Alpha and VMS. FX!32 binary code translation and emulation couldn't undo the damage DEC did to itself.

Things may change now that VMS is available on an architecture (Itanium) which has a chance of achieving critical mass. Time will tell...