Datamation ran an article about databases running on OpenVMS that never die or quit. These databases have uptimes longer than the existance of Linux and definitely longer that Microsoft product lifetimes. Here’s the article.
Old VMS Databases Never Die
Submitted by Ken Farmer 2004-11-30 Databases 21 Comments
Even though techies want to play with the latest toys, people forget what development and building systems is all about. Its about meeting some goal and producing a product that will give you revenue.. And just because WooperDooper 2.0 is out, doesn’t mean that it will give you any additional revenue. It will give you a change in systems, additional training costs, software costs, likely new bugs, and other problems though.
One system I work with (as a data feed) was put in service around 30 years ago.. perhaps a bit more. It runs all day every day on IBM Mainframes.. and is a critical system. I don’t call it legacy, because its active and in use. There is a plan to replace it in the next few years, but only to add new capabilities and to replace a suite of systems with a single consolidated one. Without those additional requirements and changes, I doubt we would be touching it anytime soon.
VMS is a great system from what i have read. In the past i have often wondered why companies maintain the older systems especially for POS and inventory. The fact is these types of systems were built for a specific purpose and were built well. Alot of these systems do not change much and would not benefit much from newer designs of the same old thing. Those that are still using VMS now have an upgrade path with HP Itanium who knows perhaps its time to revisit these older OS’s and systems.
According to wikipedia Rdb has been ported to Tru64 UNIX and Windows NT.
It seems to me that Linux and the BSDs have become too ‘mainstream’ in the last few years so geeks are looking for another OS just to show that they’re ‘different’… surprisingly the argument is exactly the same as in the Windows/Linux debate – stability. So has Linux ever been stable at all?
… Before the question gets asked, if you’d like to try out OpenVMS, might I suggest the Deathrow OpenVMS cluster. It’s free, and lets you play with OpenVMS on the VAX and Alpha. They have more nodes going in as well….
like anonymous said. try out deathrow if you have never used vms. Its a great learning spot. I use it quite often to piddle around
I tried deathrow.
Kinda scary, I actually remember how to use it. lol
I made a BBS (bulletin board system) written in VMS using only shell-like commands. In fact, it was using a z49 terminal and later vt102.
> I made a BBS (bulletin board system) written in VMS using
> only shell-like commands. In fact, it was using a z49
> terminal and later vt102.
That’s what I miss: having an integrated DB in the file system and it being successfully integrated from OS to compiler to commandline allowed increadibly powerful and flexible creation of data processing. Personally, I loved being able to do SORT /KEY=(5,2) /NODUP INPUT.DAT; SAMPLE.DAT; and instantly get a random sample of records.
Oh, about up-time? They didn’t go down unless they had faulty hardware. A correctly maintained system just didn’t go down because of software.
Ya know, I don’t care whose software or system your running. Systems that are made stable, stay stable. Whether they’re running NT, Unix, VMS, AMOS, NOS, or hand coded in 6502 assembly for your Atari 800.
What kills systems in change. Patch your OS, change your libraries, upgrade you DB, etc. All of those are prime recipes to a dead system.
Every system has bugs of somekind or another. VMS had them of course (I’ve seen them — we lost several months of work due to one), but VMS is a very mature system. There are stable NT systems out there (no really!). Obviously, we’re all familiar with Unix boxen supplying basic services for eternity until a hard drive implodes or the power supply smokes.
But once a system has been in production for several months, and gone through its major processes (like closing the year, or Christmas shipping, etc.), the systems tend to stay pretty rock solid.
These folks talking about the costs of migrating aren’t just considering the labor in moving data, but the cost of the instability that such a transition will incur.
Of course, today, we simply can’t leave well enough alone. We upgrade software constantly, most of the time blindly. And then we decry that our systems are unstable.
Forget Sun’s possibly upcoming opening of the Solaris OS code tree. I want OpenVMS opened up. It’s the one once and still popular operating system that actually is secure and still works the way it was designed to work. The only OS in existance that, in my opinion, not only has a properly designed user management system, but doesn’t have built in security circumventions out the wazoo. Seeing a port to low cost hardware for OpenVMS would make my … decade.
And yes I know about FreeVMS, but that’s not the same thing as taking the full code base of OpenVMS and porting it to other more cost effective platforms than the budget blackhole of Itanium.
I second that, OpenVMS on AMD64/Opteron would be a great combination. It would be like a marriage of old friends; Alpha CPU designers at AMD with OpenVMS running ontop 😉
Hmmmm… not sure they could do that because of gov’t contracts and stuff like that. Can they? (Open source VMS that is). I dunno…
It would be cool tho’… I would hope that if they did that they would also release their BLISS and BASIC, FORTRAN, PASCAL, ADA and other compilers for free too. Oooo! And stuff like FMS too!!
(or MSJ in my comments in the OpenVMS code I had the privilege of touching
“Ya know, I don’t care whose software or system your running. Systems that are made stable, stay stable. Whether they’re running NT, Unix, VMS, AMOS, NOS, or hand coded in 6502 assembly for your Atari 800.”
Bull… I had 112.000 perfectly stable W98 installs and by design, these would not be stable for more than 2-3 days… If you really think the only reason VMS run “for ages” is because nobody ever touches it you are one crazy man… that’s for sure. 🙂 lol
I had stable, never touched NT4 + W2Ks, these would indeed run very long, albeit not doing much more than close to nothing, anyway, but after a couple month, its over.
“not sure they could do that because of gov’t contracts and stuff like that. Can they? (Open source VMS that is).”
Little known fact that VMS has always been “open source”. The source code could be ordered from DIGITAL’s standard catalog and came on microfiche. All serious sites had the microfiche, a microfiche viewer and a copy of the Black Book in their System Manager’s office. VMS was written in BLISS, MACRO and FORTRAN.
VMS was, without a doubt, the best operating system on the market. But if you could not (afford to) run it, then UNIX was a good (but distant 30-year old) second to VMS. Similarly, X.25 VANs (which largely ran on VAXes at the end) would have been much better suited to run the global Internet on.
But we all got screwed (and I do mean screwed) by Former “Senator” Gore’s initiative that turned the NSFnet over to the public, complete with async-TCP/IP and it’s predisposition to UNIX. (You see, he actually did “invent” the Internet, in a sense.) His Act gave new life to the only UNIXesq vendor (SUN) left standing after DIGITAL had out-competed everyone else, save HP and IBM. (SUN was only still around because it shielded itself from creditors in BANKRUPTCY COURT while it switched from VAX-like 68K architecture to Alpha-like SPARC, where as DIGITAL had to PAY to develop Alpha, which kicked SPARC’s arse before Intel allegedly stole its plans to build the Pentium II/III/IV and settled-out-of-court with DIGITAL by buying the Alpha chip and shutting it down.) Worse, the birth of the UNIX-fed TCP/IP breathing Internet spelled the death of DIGITAL, which will go down in history as one of the greatest American startups (thanks to its founder Ken Olsen). And even worse, Gore’s Act has condemned us all to a world where computer viruses from the other side of the planet can take-out half of our local economy in milliseconds (because of TCP/IP’s architecture vs. X.25 centralized architecture) while also enabling our technology gems (in the form of open-source knock-offs, or even ripp-offs) to fall into the hands of people who make $5000/yr (in India). And if this weren’t bad enough, the phone companies themselves are collapsing onto an offshoot technology (VoIP) that will soon extend the power to tap all of our telephones to any good hacker.
You all forget one of the prime reasons VMS was so great to program on. RMS the journal file system that prevented beginner programmers like myself from screwing up an edited file. Or the simple linking of files compiled from differed compilers into a unified program. You could always take best of the different languages to write programs. I will always miss the Ada compiler. And I still remember that the COBOL compiler could do Math functions faster than the Fortran compiler. Or the Fortran optomizer that would great performance if you left out any output statements. Or one line calls to remote decnet systems to distribute program execution. Too many great things lost.
Preach it, brother!
Little known fact that VMS has always been “open source”.
Given its license more like shared source though.
Funny how the systems that are called “legacy” are the same ones that you can still get hardware support from the OEM on. We’ve got a couple of AlphaServer 2100s that were brand spankin’ new…in 1994…but are still doing what we need them to do and still getting hardware support from the OEM. Has anybody tried to get support from that Roundrock Texas company on hardware that’s past 5 years old?
I’ve worked in a shop where all of the VMS production clusters were available for multiple years, continuously. We could update the OS, put in the occasional patch, etc, and still have the application available regardless of the uptime of a single system.
For those not familiar with what a VMScluster is think of it as a single system image virtual machine with up to 96 (and the limit may higher now) physical hardware nodes. Each of those nodes could be a multi-cpu behemoth, or a single CPU 1U box. The nodes can be in the same room, hundreds of kilometers apart, or a combination, and still seem like one virtual machine to the application end-user.
All of this scalability, availability and fault tolerance come from the same engineering group that does the OS. This means you never have to worry about the third-party finger-pointing blame game _IF_ something goes awry.
Pity more PHB (pointy hair bosses) don’t understand how that kind of reliability can positively affect their bottom line.
VMS users tend to remember good VMS applications. Like the mentioned databases. Or the VMS compilers. And all those little widgets like CLI/DCL, RMS, DECnet etc.. How about the Polycenter products? Great stuff! But was there any VMS application released since the early 90’s that is really memorable? Let’s think VMS again!
I’ve seen several posts about being able to run OpenVMS on a Intel plateform. While it can’t be run natively, you can run a binary compatible emulator like SIMH (google it). It emulates a VAXServer 3900 (I believe), and is _binary_ compatible with “true” vaxen. I run it on my laptop. Works great, and supports networking (DECNet, TCP/IP, whatever) and clustering. Not bad if you just want to tinker with a VMS system, but can afford to buy a VAX.
Runs on Linux, Windows, etc. An old VAX doesn’t really cost that much these days in the second hand market, but don’t expect wonders. There are several free shell VMS services these days, one already named, the other known one is DECUS.
Really, try it. You’ll find advantages or things which are different than the current popular OSes.