ComputerWeekly takes a look at OpenVMS, one of the most stable OSes ever placed in production. Originally known as VMS, it’s probably the best designed and most robust general purpose operating system in existence. It’s also one of the least-known and appreciated, simply because it works quietly in the background without drama, unlike its noisier and more fussy siblings and offspring.
There’s an excellent book about the development of NT called “Showstopper”. Dave Cutler who was part of the DEC team that designed VMS headed up the original NT development
Cutler was also responsible for portions of RSX-11 and ELN prior to working on VMS (and then departing for Microsoft).
To be honest though – some of the really neat stuff has been done in the last 5-6 years and was done long after Cutler left, Galaxy (Galactic Memory – mentioned in the article) is a very neat hardware partioning scheme.
Obligatory Trivia – (and this might still be true today) VMS was/is written by a team primarily based in building 3 of the Digital Spitbrook road campus in Nashua NH. The other team in the same building (different floor) was the Digital Unix team – there was plenty of friendly rivalry between the two teams (competing banners hung around the building etc).
Andrew
DEC Employee : 1990-1997
At one Compaq released OpenVMS and Windows Integration for Dummies. Now thats a book.
There’s some interesting stuff in there. Just disappointing that the article didn’t include a link to the product they’retalking about.
Anyone can provide some links to screenshots of the OS ?
Anyone can provide some links to screenshots of the OS ?
Only on OSNews…
The way OpenVMS looks depends on the windowing system it’s running (if any). There’s no a definite ‘OpenVMS look’.
the problem is that just because a guy who was the head of the VMS team at DEC breaths life into NT does not mean NT was any good. in fact, it was probably not anywhere near where the guy thought it should have gone.
I don’t think it’s even windowing – just command line. But I haven’t seen VMS running since 1993.
You can find quite some graphical screenshots here:
http://toastytech.com/guis/DWindows2.html
And Windows is just funny: services on top of something like a microkernel, yet if one of these services dies you get a BSOD. That’s not exactly how it should be!
Windows is not a true microkernel, same with OSX. They are both hybrids between micro and mono, which seems to be the best approach for what they are used for.
The truth is, it is not efficient to message pass to modules that read and write to memory or to the disk, that stuff belongs in the kernel.
If you want to know what VMS looks like, take a UNIX shell and expand all abreviations:
f77 -> FORTRAN77
passwd -> SET PASSWORD
ls -> DIRECTORY
and so on.
VMS is as verbose as it is stable.
How open is OpenVMS?
OpenVMS is something of a silly OS choice for modern applications. If you need a high availability OS for mission critical applications, I don’t think you need to look much further than z/OS on zSeries mainframes, which will provide you with a much higher degree of support through IBM than HP is willing to offer, considering HP’s main push is to get enterprise customers on HP-UX/Itanium. Alpha and OpenVMS have been relegated to a largely “legacy support” status.
bzzzzzt!
It may have been verbose, but things like search are easier to remember than grep.
DEC from 1984-1996, all over the place, ending in OpenVMS Engineering.
It was a lot of fun, a good ride. Too bad it had to end (or is still in the process of ending. I was just a bug-fixer for OpenVMS Security (and maintained the security_server) but I really enjoyed poking around! One of my scariest bugs was logged by one of THE VMS gurus and I was a “noob” in VMS (apparently no one else wanted to look at it because of who logged it 
so I spent 2 months looking at this bug before filing a report which was that it was NOT a bug, but would become a bug if certain features that were available in the code were ever activated (they were not, of course, ever activated). Anyway, it was a lot of fun and VMS will always be my favorite OS, with BeOS being my favorite Desktop OS…
Mike
The truth is, it is not efficient to message pass to modules that read and write to memory or to the disk, that stuff belongs in the kernel.
Ah, but effecient isn’t everything, not even close. Especially in enterprise. Other factors such as reliability and security play an important role. Just look at the growth of Java/C#. It’s all about cost/benefit.
> if you want to know what VMS looks like, take a UNIX shell and expand all abreviations:
You apparently don’t know (VMS has abbreviations, UNIX does not). Most of the common commands abbreviate based on
their uniqueness (1-4 characters).
I happen to agree with Clay. In fact, this is nothing but a free HP advertisement, but also happen to dislike the proprietary nature of OpenVMS.
OpenVMS is inflexible in the sense that once you buy into the needed hardware, you probably don’t want to run anything else on it. It costs an arm and a leg and locks you into one single proprietary vendor, exactly the kind of situaton that most companies do not want to be in.
The reason some admins like OpenVMS is that they don’t want to retrain. Having said that, I am all for if it’s not broken, don’t fix it.
> The truth is, it is not efficient to message pass to
> modules that read and write to memory or to the disk,
> that stuff belongs in the kernel.
This statement isn’t entirely true, the problem with microkernels is a result of it being very expensive to switch between kernel mode and user mode. Since message passing requires kernel mode there is an increased number of switches between the two modes. In a microkernel much of what is normally performed in Kernel mode is done in Userland which is generally a safer scheme of doing things, but this increases IPC traffic which requires lots of bouts into kernel land where as a monolithic kernel is already in kernel mode so the message passing isn’t nearly as expensive. However, this does not mean that monolithic kernels are better than microkernels. There are many arguments on both sides for microkernels vs monolithic kernels. In general it is much easier to performance tune a monolithic kernel than it is to create a fast microkernel. However, this does not mean that it is not possible and in fact there is a lot of interesting research going on in the subject. I personally think that a microkernel is definately the way to go since it is more maintainable and fault tolerant (in general, this is true simply because there is less code in a microkernel thus when the processor is in kernel mode there is a smaller chance of bugs)
For more info on some intersting microkernels check out QNX, L4, and EROS.
I managed one in my old just. It’s good and stable, but there is one problem…
VMS is the BetaMax of OSs.
Most companies will find it’s cheaper and better in the long run to use the technology that is considered standard even if it is inferior when compared to the alternative.
Hi,
You can get OpenVMS almost for free by going to http://www.openvmshobbyist.com/
(You have to order CD’s, and will get a trial licence)
If you have an Alpha or a VAX you can test it yourself.
If you don’t have one of those, you can get a VAX emulator that can run OpenVMS from http://simh.trailing-edge.com/
I recommend all OS freaks to try, it’s a very different environment.
Tar
DECWindows, of which I posted a link to screenshots, really seems to be the “official” VMS windowing system:
“The 3100/38 was much quieter, smaller, did not produce so much heat (imagine sitting on your desk in the summer, when it is hot and having a VAX heating your feet 🙂 ) and had a large monitor, so I could run DECwindows for the first time! :-)”
http://fafner.dyndns.org/~ulmann/fafner/fafner.html
I use NetBSD for very similar reasons the author says OpenVMS is superior. NetBSD is very very STABLE and then some.
The only time I’ve used VMS was 1995 when I was a freshman in college. They had one lab filled with VAX-terminals used for email. I was a complete noob at the time.
I remember it was very verbose. But I still recall the system and those times with a lot of nostalgia!
FreeVMS seems to be dead…or sleeping.
A far cry from Windows’ pattern of patch/reboot/crash/reboot/patch/reboot/crash etc
I type this after installing today’s 5 new critical patches on my work PC. I am proud to say my home is Windows-free.
My linux computer is stable, I say. People who know would object and ask, does you computer get any action? No not really, I have to say. So, in fact I don’t know how stable my (linux) computer is.
Is it possible that there is a version of this scenario w.r.t. to applications and OS:es. Few applications reduce the probability of finding weakneses in an OS so the perceived stability will overestimate the actual stablity if compared on equal footing with other OS:es.
This is only a thought really, I’m not trying to argue about the stability of (open)VMS. I can’t say I know anything about it. I’ve used it occasionally.
/jarek
NetBSD is great, indeed, I’m running it now in fact. Yet while I have not managed to crash the betas of 2.0, I have experienced several (!) kernel crashes with 1.6, mainly when using the USB ZIP drive, but also because of my buggy soundcard.
Now I have found some places to get a free OpenVMS account, but where is any tutorial on the command line of it? I can’t figure out how those $foo:bar[bla.blup] things work…
What exactly is entailed in the switched between kernel mode and userland? Considering most bottlenecks are I/O would that mean that the faster CPUs can afford more switches?
I read it a while back and remember that I thought it was very good. Now I only remember one thing from it. That’s where Cutler comes storming out to the printer and asks the new young hire next to the printer: “Are you the one that f….. up the printer”. Apparently also a shortfuse.
http://www.amd.com/us-en/assets/content_type/DownloadableAssets/Mic…
The OpenVMS books by Ruth E. Goldenberg are legendary and the only option to learn about the internals AFAIK. And every chapter starts with a proverb.
If I had to characterize OpenVMS in one word I say polished. Unlike Windows e.g. let’s get it out of the door and deal with the problems later.
OpenVMS is open source – you can buy it. But not for modify and rebuild AFAIK. If it was and if it was also cheap or free and if it also ran on x86 then I think it could still have a renaisaince.
But asis it has become a mid-sized dinosaur. Why would anybody choose it over say Linux? Reliability? x86 is not that bad. But if not good enough then do as Google does – build the hardware from a string, a clip and a piece of gum but ensure that failover works 100% all the time.
Why is HP porting to Itanium and not (also) x86? Isn’t one such failed escapade e.g. the Alpha enough? Allthough the Alpha has been characterized as the best CPU of all times.
And DEC’s Alpha engineers now work on Itanium at Intel.
Online tutorials and documentation
http://www.pottsoft.com/home/vms/vms_tutorial.html
http://h71000.www7.hp.com/doc/73final/9996/9996PRO.HTML
Operating systems handbook, chapter 3 (PDF to download).
http://www.snee.com/bob/opsys.html
I think it’s better to begin with the PDF.
You may found other ressources on the VMS webring.
No thanks, I use OpenVMS every day of the week at work. Can’t wait to escape it into something more modern and, well, just inviting.
There are some scenarios when the google-way doesn`t work.
when a node fails you at first lose the data that has already benn processed. so you have to get the initial data-package wich was send to the node out of some archive (be it distributet like google`s or something centralised). After that you have to asign this job to another node and wait for the results (and pray that it doesn’t fail like the last one).
All this costs time, not much, but it does.
lets say such a failed node causes a delay of 1 second (i’m sure it would be much more) and your process needs a respond-time of 0.something your out of luck.
thats where those “dinosaurs” enter the game.
“Isn’t one such failed escapade e.g. the Alpha enough?”
Not alpha but HPaq is the failor. And they are doing ther best to fail again.
It would be niced if it were opensourced, however, being realistic, even if it were, how many people out there know BLISS and MACRO, let alone a required amount of obscure assembly?
What I think would be nicer would be for Tru64 to be opensourced, yes, I know, there are screams of “why not OpenVMS”, the fact is, one has to be practical, there is no use opensourcing an operating system if the netresult is an opensourcing operating system just sitting there with with done to it.
Tru64, apart from its pig-ugly GUI, is quite an interesing OS, Mach 2.5 based, nicely implemeneted n:m threading, UNIX 95 compliant. With that being said, it goes back to what I said a few years ago that it would be nice that if a product is going to be taken off the market completely, as with the case of OS/2 or Tru64, it would be great if there were a “source junk yard” where the source code can be uploaded to so that coders can experiment and continuing working on it.
Some of you seem to be wanting more information on OpenVMS. May I recommend:
http://www.hp.com/go/openvms
http://www.OpenVMS.org
Ken
OpenVMS is really enterprise grade technology, it’s not something you hack around with over the weekend, although there’s not really much stopping you from getting a used DEC workstation and a free hobbyist license:
http://www.openvmshobbyist.com
OpenVMS is literally used in several of the world’s largest stock exchanges, healthcare, government, military, manufacturing, schools, private/public businesses. In fact, I hear that Intel’s assembly lines use OpenVMS. Uptimes are measured in years, not days or months, or hours. Its reliability is unmatched, I heard a couple of good comments about stability from some of the user posts. My linux and mac os x boxes also sit happily with large uptimes, but I’m not really stressing the systems. For the OpenVMS systems that I’ve worked with, they could do hundreds of transactions with a database and multiple users while maintaining great stability. You almost never have to reboot, the catch being: if you’re in a cluster, which most of their enterprise customers are.
It’s not open as in open source, the open part of vms is just a horrible naming convention VMS received way before the open source movement started.
Perhaps when Itanium gets around the price of Xeons in 2005 will you see an everyday version of VMS for your personal needs.
I also want to mention that they’ve ported bash to openvms which can be found at: http://sourceforge.net/projects/gnv
but here are some commands you can play with
set default ~ cd
show users ~ users
create directory ~ mkdir
show system ~ ps
of course you can abbreviate most things by using 3 characters so
instead of set default you can type set def.
as to what openvms looks like, it uses CDE as its default window system.
the command prompt looks like:
$
directories look like:
dka0:[username.home]
And OpenVMS runs modern stuff too like Java, WebLogic, NetBeans, etc.
A nice ms windows integration tool can be found here:
http://www.openvms.compaq.com/commercial/bridgeworks
download it, it runs on win2000/NT/XP and is free.
When I think of VMS I think of the Vax machines that where used at the community college I wen’t to that are just consoles with orange text. I hadn’t actually ever seen it with any kind of windowing system except for the screenshots from that one archive.
And DEC’s Alpha engineers now work on Itanium at Intel.
Nah, they worked on building the Opteron.
http://zdnet.com.com/2100-1103-995915.html
I found this overview of who’s who and CPUs
http://www.cs.clemson.edu/~mark/architects.html
Interestingly it shows that Cutler also was involved in the hardware. Which may be why both the VAX and Alpha has features such as interlocked doubly linked list support in hardware (PAL code).
Sorry, not to troll and go off topic and not to detract from OpenVMS — however I was merely mentioning that if one was looking for an extremely stable operating system — that runs like as stable and humming as an appliance like a refridgerator — NetBSD is like that. OpenVMS is too. They are different OS’s but they at least share that trait in common. The basic nature of humans is chicanery so that’s unfortunate that your kernel crashed (*cough* bullshit* *cough*). It may segfault but only through your own stupidity when rolling your own kernel so to speak not on GENERIC on -stable.
As far as security goes.. well OpenVMS might have Net slightly beaten through obscurity and other reasons — however, Net has fewer bug exploits on bugraq then even OpenBSD. It depends what your using it for really.
However, the propolice stack patches run on Net now and it’s VERY VERY easy to checkout the OpenBSD crypto code through CVS and merge what Itojun didn’t into the NetBSD system — since OpenBSD and NetBSD have the same code base.
So you can have the same stability of and security of OpenVMS if not more (particularly security if you merge some Open code base) with NetBSD with a little expert effort.
Off the record though : I’d like to see both plan 9 and NetBSD be risen from the graves before OpenVMS is. Technically, Net isn’t in the grave though — but you know what I mean.
Sorry, for the rant. Carry on brave OpenVMS connoisseurs/hobbyists.
*typos:
bugraq= bugtraq http://www.bugtraq.org/
I’m sure I used “your” instead of “you’re” in one context too– oh well such is life when you are half awake hacking in the middle of the night when most people are sleeping
Sorry, for slaughtering the thread guys ! Carry on by all means and ignore my posts if necessary.
Clustering ?
http://web.mit.edu/netbsd/www/whatis.html
ttp://www.myri.com/news/96a25.html
http://www.cs.bell-labs.com/plan9dist/
http://www.ccrc.wustl.edu/~milind/vita/milind_vita.new.html
Ok, Stability and superior protability ? NetBSD.
Security ?
Parse the NetBSD base into cylcone or just be happy with the propolice patches and merge some OpenBSD crypto.
http://www.research.att.com/projects/cyclone/
http://www.openbsd.org
Security and Stability ? NetBSD/OpenBSD hybrid hacker tailoring.
OpenVMS ? Obsolete.
http://www.myri.com/news/96a25.html
BTW, I realize that I linked to the deprecated athena/NFS url page at MIT. NFS needs to be phased out for something superior. NFS really is ancient technology. Then again OpenVMS is too. Hackers have tackled the easy things in the BSD world but have yet to take on the big projects — namely replacing NFS with an entirely new concept.
It was a GENERIC kernel. And a GENERIC kernel crashing every time (!) I put some load on a USB ZIP drive, and of which the authors admit that it crashes when the system runs out of memory, does not appear very reliable to me.
Yet as I said, I believe 2.0 is going to be great. I have not been able to crash the betas yet (except for the soundcard, but that’s not NetBSD’s fault).
—
Anyways, Manik, thanks for pointing me to that PDF file! Quickly looking through it, I have already seen some advantages of OpenVMS compared to UNIX (resuming a disconnected session without “screen”, versioned files à la NetWare).
clustering
/usr/pkgsrc/parallel/pvm3]% cat DESCR [19:42]
PVM (Parallel Virtual Machine) is a portable message-passing programming
system, designed to link separate host machines to form a “virtual
machine” which is a single, manageable computing resource.
The virtual machine can be composed of hosts of varying types, in
physically remote locations. PVM applications can be composed of any
number of separate processes, or components, written in a mixture of C,
C++ and Fortran. The system is portable to a wide variety of
architectures, including workstations, multiprocessors, supercomputers and
PCs.
PVM is a byproduct of ongoing research at several institutions, and is
made available to the public free of charge
Kaiwai wrote: It would be niced if it were opensourced, however, being realistic, even if it were, how many people out there know BLISS and MACRO, let alone a required amount of obscure assembly?
When a consulting client bought a Bliss license, there was the Reference Manual for strict definitions, but the best tutorial available was the VMS source listings (on Microfiche at that time; since moved to CDROM). Having such a large body of working code to read was a great example at least of how one programming shop used the language. (To this date, VMS does not make much use of the Bliss FIELD feature.)
At any rate, the best programmers in any language not only know that language well but also know other languages. I rarely write code in Bliss anymore, but knowing Bliss is quite important (especially if I want to see how some part of VMS works “under the covers”) even today.
Tar wrote: You can get OpenVMS almost for free by going to http://www.openvmshobbyist.com/ (You have to order CD’s, and will get a trial licence)
Actually totally free, since VMS Hobbyists are also allowed to borrow the media from a friend (including employer, if the boss says ok). Having friends is good, for learning the easiest way to do things etc. In years past this was mainly done through local DECUS meetings, but in these days of the Internet try:
NNTP: comp.os.vms
Telnet: eisner.decus.org
Clay wrote: OpenVMS is inflexible in the sense that once you buy into the needed hardware, you probably don’t want to run anything else on it.
Sometimes, not switching operating systems is good. Would you like to have the Automated Teller Machine reject your transaction saying “sorry, we switched from VM to running ACP for this shift to do some other work” ? VMS is really aimed at situations where downtime is unacceptable, rather than just inconvenient.
For developers who are not buying into the whole Uninterruptible Power Supply way of life, there are cheap older VMS systems available on eBay (both VAX and Alpha). Sure it would impress my friends to have a 128 processor GS1280, but when it comes right down to it, regardless of operating system, how fast can I single step in the debugger ?
The stability of OpenVMS is awesome – yes UNIX is good but VMS really is better. We had some old DEC boxes at my previous job – as well as shiny new dark blue alpha servers. One of the old ones (not sure how old – DEC badged beige desktop sized thing) was used by all us developers every day. Hardware failure got it in the end – drive crashed. Thing is it was soooo long since anyone there had needed physical access to the machine no one knew where the box was! Took a bit of hunting and on opening the thing it had about an inch of dust across the whole circuit board. Continous uptime of at least 5 years. Probably good for another five with its new HDD and dusting.
VMS a silly choice for modern applications? A lot of banks don’t think so. What’s really funny is it still consistently outperforms an equivalent IBM product.
I’d agree with you about the IBM support though. IBM products requires a LOT of it. There is probably more code in PTF’s for one IBM platform than there is for the whole of VMS.
So you can have the same stability of and security of OpenVMS if not more (particularly security if you merge some Open code base) with NetBSD with a little expert effort.
Having run NetBSD and OpenVMS on some machines, I noticed that NetBSD was quite slower, though. Some might see this as a detail, but stock traders and hospitals might have another opinion 😉
Sorry but I respectfully disagree with the idea that VMS is verbose.
It is true that you can spell out commands explicitly but noone does that in the real world.
Any command can be abbreviated down to as little as 3 letters as long as they remain unique commands.
DIRECORY =DIR
DELETE =DEL
RENAME =REN
Etc.
The word Open was added because VMS originally conformed to POSIX, not because it was open-source.
One thing I love about VMS is its ability to have multiple versions of a file ( same name, same location in directory).
Something that nobody else does as far as I know.
Every time you edit a file, a newer version is created while retaining the original.
A file would look like test.log;1.
As you edit it, the number increments so that test.log;1 would become test.log;2, etc.
A file can be versioned all the way up to test.log;32767.
This capability allows you to roll back to a prior version if bugs in code are found. No muss, no fuss.
I’ve been saved on more than one occassion by this.
How “Open”?
There is a project called the GNV, Gnu’s Not VMS, which is about bringing UNIX open source to VMS, which includes fully emulating a UNIX shell, file system, all under the umbrella of VMS. Openoffice is being ported currently to VMS using this, see http://www.oooovms.dyndns.org/openoffice/
The big issue is that people are sat around using version 6.2, which is like using a PC with Windows 95. If you want to be in the W2003 space, you need 7.3-2, and you can play with IPv6, SSH/SSL Kerberos, Java, Fibrechannel, Dynamic volume expansion (now that a trick I’d like to see other o/s manage on line without a reboot!) And I’ve barely scraped the surface.
Someone also said in the comments that it is “propietary” in terms of hardware. So just how stable do you think VMS would be, with some 2-bit PCI card that bent the rules to suit, plus a buggy driver, in it?
And it can be cheaper, one VMS box can run many apps, rather than one box one app.
You wanted to know how “Open” VMS is? Many of the industry-standard development tools and applications are available now. Add to this OpenVMS’s fault tolerance and security, and you get a very powerful Ebusiness tool!
Ok, here are a few of the “Open” parts that are actively supported on OpenVMS:
Open Source Tools – Ported to HP OpenVMS
» OpenVMS freeware
» open source tools available for download
» GnuPG
» GNV
» GTK+
» libIDL
» Mozilla
» Perl
» Stunnel
» VMSTar
» ZIP
» source code kits
» CD Record
» CDSA
» Kerberos
» SSL
» other open source tools
» MySQL
» Python
» OpenVMS e-business products
Web Browsers
» Secure Web Browser for OpenVMS Alpha(based on Mozilla)
» Mozilla
» LYNX
» Netscape Navigator V3.03 for OpenVMS Alpha
» Netscape Navigator V3.03 for OpenVMS VAX
Web Servers
» Secure Web Server for OpenVMS Alpha (based on Apache)
» PHP
» Perl
» Tomcat (JSP)
» OSU HTTP server
» CERN HTTPD server
» WASD VMS Hypertext Services
Application Development
» NetBeans
» XML Technology
Java™ Platform
» Java for Alpha Systems (SDK, RTE, Fast VM)
» Optimizing Java™ Technology Software Performance on HP OpenVMS — Tips and Tricks for Users
GNV
GNV (GNU’s Not VMS) is an open source, GNU-based, UNIX environment for OpenVMS that provides UNIX application developers, system managers, and users a UNIX-style environment on OpenVMS. This facilitates development and porting of UNIX software to OpenVMS. (GNU is a UNIX-like operating system that is free software.) GNV provides a UNIX-like shell (command-line interpreter) environment and a C Run-Time Library (CRTL) supplemental library to provide utilities typically found on UNIX systems. The shell used by GNV is Bash (Bourne-Again SHell), from GNU, using the POSIX.2 specification.
GNV runs on OpenVMS Alpha Version 7.3-1 or higher.
can anyone give a link to the mysql port to vms?
Also yeah, I love the file versioning, it’s great!
if you’ve got a bunch of versions of one file, you type the command “purge” to get rid of all but the latest one. Sometimes I catch myself typing “purge” in mac os x by accident.
Also, VMS is going under some government certification to make it more UNIX compliant so it can be adopted by government. This is what caused the need for GNV and other projects at HP.
VMS is growing and perhaps someday it may even come to your laptop natively.
Also, from my use of the java fast VM on alpha, it seems to blow other systems out of the water in terms of speed. I tested against x86 and powerpc implementations, not sun, etc.
There are a few public-access OpenVMS systems out there, although these are run by hobbyists and might not be as reliable as a “production” machine. This includes the tongue-in-cheek-named “Deathrow Cluster”. Which can be found at
http://www.openvms-rocks.com
Yes, when you sign up there you do indeed get [email protected] as a mail address.
Bascule wrote: “You need a high availability OS for mission critical applications, I don’t think you need to look much further than z/OS on zSeries mainframes, which will provide you with a much higher degree of support through IBM than HP is willing to offer.”
z/OS on zSeries???
Higher degree of support???
Get real. We have about 30 VAXen and Alphas and two zSeries machines between 6 sites. VMS software and hardware support is just as good as what we get from IBM, if not better. The main difference is that the VMS systems are up 365/24, the zSeries still get their weekly “Therapeutic” reboots.
“OpenVMS is something of a silly OS choice for modern applications” … another guy said. It sounds like he’s talking about the calculator application on Windows, because I’d like to point out that “modern” applications of signficant size often rely on mulitple computers working together – like a database server, and web server, and a middle layer, etc. For “modern’ applications like these, no one with a mind larger than a calculator application would place the database server on Windows.
I’ve been running a public access OpenVMS cluster for a while now. It’s security and clustering can’t be beat. Anyways, for anyone wanting to play and poke around a OpenVMS system, check out:
http://deathrow.vistech.net or http://www.openvms-rocks.com.
Actually, these two links technically go to the same machine. Check it out.
Have you seen the title of their page?
“OpenVMS.org – The OS with uptimes longer than MS Windows support policies”
I just love the people who assume because they haven’t heard anything about OPENVMS that it has been “relegated to legacy support” It is in continuous development, and has an excellent support team. Ask anyone who supports an Enterprise high availability system on OPENVMS about how long it takes to get a problem resolved by HP OPENVMS support. It usually takes one phone call. The techs are knowledgeable and understand the urgency of downtime. Just talk to IT managers in Power plants, banks, hospitals if they plan on moving off of OPENVMS to some flavor of Unix and you will get a resounding “Are you crazy?”