Stories with similar links:
- OpenVMS/x86_64: Proof Points to x86-64 First Boot (September 14, 2018) via trn 4 years ago | 11 points | 16 comments
- OpenVMS/x86_64: Proof Points to x86-64 First Boot (May 29th 2018) authored by trn 5 years ago | 12 points | 2 comments
My own VMS history:
My own exposure to VMS was long before before any exposure to UNIX - partly why VMS and DCL still just “feels right” to me.
This was back in days when I was using a Commodore 64 with GEOS (with either a 300 or 1200 baud modem! [got a 2400 baud Practical Peripherals brand Hayes clone later]) as my home computer system. Mainly purchased for gaming and productivity, I learned BASIC and later 6502 assembly with it.
I started using VMS via a dial-up account. The account was owned by the older brother of a good friend who attended the local university, and he didn’t mind sharing it as long as we, and by extension, he, didn’t get into any trouble.
This was the first real and networked system that I had access to - a large VMScluster. It was also my first exposure to the Internet back in it’s early days. I was able to hold onto that account for quite a long time. A little later during this same timeframe (and connecting via the VAX), I was able to get access on two other networked (non-BBS) systems - PRIMOS and Wang VS. But VMS was “home”.
Fast forward some years, and I’m a computer science student at the same university, which now hosts a greatly expanded VMScluster (one of the largest in this part of the country as I recall). Computer science (CSE) was part of the engineering department (hence the E in CSE), and we had (essentially only) UNIX (Sun SPARC) systems to use.
These were not good Sun systems. Or, if they were “good”, they were still terrible. I mean, go ahead, try to get real work done on an IPC’s and IPX’s - usually running CDE, and swapping over NFS. Even switching to twm didn’t make these much more usable. There was a large server you could connect to but it seems it was always overloaded by students and faculty, and you often had better luck trying to find an idle workstation during the busiest times.
Ocean Engineering (OE) students had NeXT systems (way nicer than the Sun systems). We would sneak in to play with them sometimes, but we weren’t really allowed to. (They were nice enough that I later bought a NeXTstation Turbo for use at home, when I could afford it, which was the first real UNIX system I ever owned, but I digress.)
At the school, the business technology department (IRM - Information Resource Management) was not considered to be part of the engineering department. These business students used the “real systems” (VMS) for their work. The VMS cluster also ran all the important university software, was the main network gateway and server for the school - pretty much everything ran on VMS. All students could use the VMScluster, but most did only for access to their e-mail with PINE and never really used more than a few basic commands.
Of course, I had a fondness for the VMS systems, but it was practical as well. It almost always faster, easier to use, better debugging environment, not to mention less overloaded than the available Solaris systems. The DEC C and other compilers were a lot nicer to use than what was on the Sun systems. I did almost all my real work on the VMS system, and the Sun systems were just X terminals open dtterm and run telnet for me.
I can’t remember what it replaced, but I needed a new computer, and got a great deal on a hefty and powerful (used) VAXstation for my use at home, so I went to the IRM offices to met the System Admins of the VMS systems. I was tickled to see they were all also using X on their desktops - but running VMS (DECwindows/Motif CDE). They explained that the university educational license applied to students, and had me bring the VAXstation in, connected it to the network, and installed VMS on it for me with all the layered products (ALL-IN-1!) under the university license (which if I remember, I had to renew with them yearly). I had also (through other channels) obtained WordPerfect and Lotus 1-2-3.
Fast forward again some years (I wasn’t there at the time), and I heard that they merged IRM and CSE departments into a single unified computer department, and VMS was the loser in the deal, and everything migrated to Solaris, and many hearts were broken it was wasn’t exactly a smooth transition. Today, sadly, VMS is no longer there.
Today, I have the ultimate soft spot for VMS, because it was always the environment that was the most pleasant to work with, was always available, and always worked right. In stark contrast, using UNIX was almost always a (very) negative experience, even when I had access to better machines, and it was never the environment that I was most comfortable with. UNIX was a toy for the geeks, and VMS was a system for real work.
It’s my opinion, at least back in the day, that unless you really prided yourself in being one of those computer geeks (and maybe I was and probably still am, even if it isn’t my primary passion), you avoided UNIX, and I couldn’t see any good reason why anyone would ever pick UNIX over VMS. Later, the only good reasons to pick UNIX over VMS were ubiquity and cost - which are two very good reasons indeed.
If I could practicably use a VMS desktop (even if it had to be DECwindows/Motif!) as my main desktop system, I certainly would (though I understand that VMSI is focusing on servers only, at least to start, and the last supported VMS desktop configurations are now quite aged.) I have access to an IA64 VMS system, own an old and slow Alpha capable of running VMS, and run both Alpha and VAX VMS in emulation, but sadly can’t say that it’s my daily driver anymore. Even now, almost never does a week go by using UNIX that I don’t think about how much better/easier/cleaner it would be if things were ‘the VMS way’.
I’m certainly going to call VMSI soon to see what my options are for running it at home, even if I pay to pay for it.
A little side story.
This was only a local rumor, but, so the story goes, a group students (and mostly ex-students) were using VMS illicitly under the university educational license as a platform for a very-much-not-educational commercial venture, and were shocked by the costs of licensing VMS with all the layered products - which back then was often quite high.
To get around it, they had a student maintain his enrollment for many years, since it made “business sense” for them to keep him enrolled rather than pay the VMS licensing fee.
I would hope that they extend the existent hobbyist program to the x86 version. It’d make a great way to introduce people to the platform, especially if companies using it have issues attracting talent who can work with it.
Please make this a blogpost. It’s worthy of being one. You can just copy-paste this comment.
I would very much like to run OpenVMS on my laptop without emulation.
Ive seen you saying this on a few threads, even talking about buying it. I cant remember if you told us about your VMS experiences or what you missed about it. Im definitely interested.
OpenVMS was never my primary operating system, but it was what I used for a lot of school assignments in college (we had a VAX and VT220’s scattered around campus). My Operating Systems course also did a deep dive on the design of VMS.
I actually got into that VAX before I got into that college. When I was young, around 13, I wanted to get on the internet. At the time, the internet was for major businesses, universities, and government entities…the first commercial ISP wasn’t even two years old yet.
I grew up in a small town that had a satellite campus for that college. Being young and naive, I called them up and asked if I could use their internet connection. They must’ve assumed I was a student or something and gave me the dial-in number for a terminal concentrator in my hometown. I could dial in there (thanks to the satellite campus) without long-distance charges and then telnet from there over to the VAX at the main campus.
Of course I didn’t have an account or anything so that didn’t work. I quickly discovered, however, that they didn’t limit where I could telnet to from that local terminal concentrator. I immediately started exploring the internet, especially the early talkers (I was a regular, and by far the youngest regular user for a long time, on Foothills).
I got caught by the admin only after months. It was a simpler time back then, and in exchange for telling him how I bypassed the non-existent security, he just gave me a “test account” on the university system. I used that until commercial internet service became available in my town a couple of years later. All of my early Internet adventures were through a DCL command line inside a Terminal screen on the Amiga.
So, obviously, I have quite the soft spot for VMS. It figured very much into making me who I am today.
(I remember @quad telling a similar story in some other thread: the same sort of security bypass by discovering that the dial-in terminal concentrator didn’t limit where they could telnet, etc. It was the first time I’d ever heard someone else have a similar story, and, at least for me, it was a real moment of…identification? camaraderie? I dunno. It sounds silly, but it really stuck with me because I had never met anyone with that exact experience before and it’s such a specific and random thing to have happen.)
Neat way to start! Thanks for sharing it.
I didn’t (couldn’t!) link to the official press release, because the page (/news.html) doesn’t have a way to link to a specific news article, and didn’t show the interesting technical detail, which is the actual output from their “debugging session that achieved the First Boot criteria”, but here it is, in case anyone is reading here and lobste.rs and doesn’t want to poke around the VMSI site to find the roadmap to the eventual release.
Cool!
I’ve never used VMS. What would it be good for in 2019 on an x86_64 machine?
@trn’s description is great. I”ll add two things:
Security is probably behind BSD’s and Linux’s now if you’re using all the state of the art sandboxes, compiler tools, and so on. What any given company or distro does varies considerably. VMS’s default capabilities (circa 1999) were better than UNIX’s defaults, though. It had finer-grain permissioning of processes, CPU/memory metering like in clouds, and most services off-by-default like in OpenBSD. Since it was expensive and unfamiliar to hackers, it also had only about twenty something CVE’s versus hundreds for Linux and Windows. Better security design plus obfuscation equaled less attacks in practice. Note that higher-security extensions like SEVMS and VAX Secure VMM (pdf) existed. Just nobody buying higher security stuff at the time. VMM was cancelled. Idk if SEVMS still available.
I don’t have any market data about banking use. I do have a banking example (pdf) that, on 9/11, perfectly illustrated why many companies run critical stuff on VMS clusters. It also shows off the AlphaServer hardware that kept running as all the Wintel machines crashed. Just mentally compare that to all the outages you see on these online services running on Linux. That kind of thing isn’t tolerated in OpenVMS shops. Cost a lot to get the reliability, though.
HP had a guide on using OpenVMS for disaster tolerance. It goes into a lot of detail about the topic in general that can aid efforts outside of OpenVMS. It also shows how thoroughly the OpenVMS ecosystem had this covered.
That’s the problem with a proprietary monolithic operating system that is designed and not “grown” organically - it stagnates once the stewards begin to neglect it.
OpenVMS did not get the attention it deserved after passing through so many hands after DEC (at Compaq and HP/HPe).
VMSI’s team consists of numerous ex-DEC employees and many original core VMS developers. Their senior engineer who lead the x86-64 effort is the man who oversaw the VAX to Alpha transition at Digital.
I really hope they can succeed with VMS.
I agree. There might have been extra motivation for letting it stagnate when they had two, competing products with NonStop being more profitable. I like that the new team has the necessary experience. What they’ll do for security… who knows.
VMS excels where you need fine grained security, tight clustering, functional yet relatively simple management of a large cluster, unified and simple networking and communication between users and processes, most importantly, extremely high reliability and stability, and the ability to support a very large number of interactive users or terminals on modest hardware.
Historically, VMS was also rather unique because it was a single software system and management platform which scaled from very modest pizzabox form-factor desktop systems to extremely large mainframe-class clustered systems that spanned entire raised-floor datacenters. That may be taken for granted now with large UNIX and NT systems, but remember that VMS was a mature product when UNIX was still in it’s (commercial) infancy, and the UNIX market and UNIX implementations were very fragmented, with the datacenter-ready UNIX systems all being very different independent developments.
VMS systems were, and still are, used in safety and life-critical applications in industries like defense, oil and natural gas, power generation, nuclear, shipping and transportation, rail, factory and process automation, etc.
(While some big finance/banking users existed, as far as I know, VMS never really took off in that market, which was, and still is, dominated by IBM, Stratus, and NonStop.)
There used to be a very large community of scientific and research users (both in academia and industry) who used VMS, mainly in high-energy particle physics and astronomy. Most of those users used the computers not because they were “computer people”, but because they just needed computers in the course of their work and wanted systems that would “just work” and otherwise stay out of the way.
Somewhat straying from the original question, but, remember, was also a time not so long ago, that UNIX vendors - especially supercomputer manufacturers - were aggressively marketing VMS compatible and interoperable systems.
Convex was big player with CoVue - The Convex-to-VMS User Environment which offered system (DAP, DECnet), user (DCL, EDT), and developer (VMS RTL) compatibility with VMS.
UNIX was a tough sell in traditionally VMS-heavy markets, and an impossible without getting the users to buy in. There were very real management concerns about retraining staff and lower productivity — even if the replacement hardware was more powerful, it’s useless without users.
Good thing they are keeping it isolated and far from wide spread introduction. It’ll be amusing how many kilo dollars per core its going to be to have the honour of this dead end zombie