Legacy Systems: Tried and True Systems Whose Time Has Come
December 13th, 2013 By: Industry Perspectives
Duane Harris is CEO of Nemonix Engineering.DUANE HARRIS
In an era of technology turmoil—with news of security breaches, overloaded servers and major corporate and government computer failures headlining the front-pages—there are unexpected islands of calm populated by extremely stable legacy hardware systems. Ironically, some of these oldest citizens of the data center also happen to be the most stable and secure. The extended value these legacy systems still generate in modern computing is a story worth noting—and perhaps learning from.
These servers are still some of the most reliable, secure and indestructible systems in the data center today. At the top of this list of end-of-life machines is hardware running OpenVMS, an operating system built by Digital Equipment Corporation back in 1977, and updated by HP ever since. New and old versions of OpenVMS still run mission-critical applications on legacy hardware, as well as modern, Intel Itanium-based hardware manufactured by HP for some of the biggest names in government and industry.
Below are some of the benefits of OpenVMS that other systems have yet to match:
- Disaster Recovery. OpenVMS’ fault tolerance and disaster recovery features are legend. During the 9/11 tragedy, a major international bank with North American headquarters, located less than 100 yards from the World Trade Center was among a mere handful of companies that remained online—primarily due to its reliance on an OpenVMS-based disaster recovery strategy. The intense heat in that international bank’s New York data center crashed all but OpenVMS-based AlphaServer hardware. The Alphas used server clustering and hard drive volume shadowing to keep the bank’s primary system running off of drives located 30 miles away.
- 100 Percent Uptime. For enterprises requiring 100% uptime, there are few industrial-strength operating systems that can keep up with OpenVMS. For example, one of the world’s largest defense contractors has been using OpenVMS to track missile sites across the world for more than 30 years. The organization has no plans on changing either the operating system or the legacy hardware it runs on. Staying operational is so mission critical that any downtime could be disastrous, and any risk of downtime due to a new platform migration is intolerable. Many of Nemonix’ own OpenVMS customers, in both industry and government, refuse to move from OpenVMS because it simply works. And, it works with minimal intervention and few if any patch requirements. Some of our customers, such as a major U.S.-based chemical company, have commenced the migration to Windows, only to discover that the newer systems are not necessarily more stable ones. In fact, downtime instances have risen sharply in comparison to OpenVMS.
- Low Cost of Ownership. Some users stay with legacy platforms because the cost in dollars and in lost production is too high. For example, nuclear power plants must comply with very tight regulatory requirements due to the potential risk of catastrophic loss of life and property. Regulations require that if any core system hardware is changed, the entire plant must be recertified, not just the new hardware. Plant recertification could cost millions of dollars, in addition to the relatively small cost of the hardware. Similarly, in other commercial enterprises, the cost of new hardware and software is a fraction of the overall ripple effect to business processes, ancillary software licensing, retraining, retesting, recertification, and production down time. Changing to a new platform often has a huge cost footprint, far exceeding the actual cost of the system itself. Additionally, cost is reduced through the more effective management requirements of OpenVMS-based systems. A study by Wipro showed that the costs to manage an environment with 40 servers and 10 database servers were reduced by half on VMS-based systems.
- Stellar Security. Perhaps OpenVMS’ greatest claim to fame is its completely stellar security record. OpenVMS systems provide a level of security that is unmatched in the industry. According to the same Wipro study cited previously, OpenVMS is ten times more secure than other popular operating systems available today,* and has 75-to-91 times fewer unaddressed security vulnerabilities on any given day. Additionally, while many of today’s popular anti-virus programs for newer, non-VMS systems address most infections, they are not 100% effective. Additional vulnerabilities in these newer systems continue to be discovered, potentially enabling malicious compromise or re-infection before security software publishers have a chance to update their virus/malware filters.
In the great rush to upgrade to the latest greatest, it is worth taking note of the value that can be leveraged in your existing legacy hardware/software installations. Often, the business process surrounding the legacy platform is extremely valuable, representing many years of fine-tuning. It is worth pausing to save what is valuable before dismantling the entire architecture. The following are tactics that companies can leverage to extend the value of their legacy systems, and, in some instances, upgrade performance at the same time:
- A hardware system at the end of its factory life cycle is not necessarily at the end of its usefulness to the enterprise. There are technology refresh solutions that can add up to 10 years of productivity back into legacy hardware. Unlike a refurbish, where only apparent problems are solved, refresh addresses the complete system, replacing every known failure point in the hardware and then putting these extremely reliable computer systems under new one year factory warranty, extendable to 10 years. Some of these failure points include fans, batteries, power supplies, electrolytics, etc. You end up with “like new” hardware with up to 10 years of additional warrantied productivity.
- If the hardware must go, consider keeping the software platform in place. The pressure to upgrade hardware is understandable. Legacy hardware faces diminished parts supplies and a shrinking pool of technicians familiar with the aging platform. However, if you run mission-critical systems that rely on the stability and security of the OpenVMS software platform, you don’t want to introduce new risks of downtime or security breaches by migrating to an entirely new software platform. Fortunately, you can migrate to x86 and use emulation software to virtualize your legacy hardware.
- The benefits to emulation can be substantial:
- New hardware without the cost of new software licenses, other than the emulation package itself.
- Lower risk of downtime by keeping existing software applications in place.
- No expensive retraining required—your business processes are effectively unaffected.
Even though we live in a throwaway culture, where the next smartphone is just six months around the corner, the reality is that many of our mission-critical applications—from aerospace and defense, to energy and government—are still hosted, and will be hosted into the foreseeable future, by legacy computers. Enhancing the value of these stalwart systems, whether through refresh or emulation, only increases the efficiency, productivity, and stability of the overall enterprise.
Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library..
Legacy Systems are systems that are proven to work.
Despite the love of the new that pervades the IT Industry, the reality is that the important processing happens on proven systems.
A system that simply works all the time tends to be neglected until the rare occasion something breaks.
OpenVMS systems can be enhanced with new interfaces and continue to provide a stable core to a corporation’s IT operation for many years over which the fashionable systems will come and go.
I have worked on development and support of applications running on VMS which run 24x7x365 needing little maintenance, highly fault tolerant and do everything they are supposed to do. VMS is just too good to be sent off to the scrap heap, it is an OS worthy running with contemporary, state-of-the-art systems. (These systems might be considered LEADING edge, but are also seen as BLEEDING edge.) Keep VMS going strong.
Kerry FarneyPosted December 17th, 2013
I’ve been working with OpenVMS for over 20 years and there is no other operating system I’d rather work with. We have Windows Servers in production environments (we’re a manufacturer) and they’ve been ok, but no where near as stable as our tried and true OpenVMS systems!
David ElinsPosted December 17th, 2013
In my humble opinion what VMS needs is a massive user interface update. The internals are just fine – in fact (for you techies) I’ve always been surprised that no other OS picked up the idea of logical names.
There are grass roots projects to port it to other architectures – it would be nice to see HP take that effort up.
(Disclaimer – I was a DEC employee and did a minor part of the work of porting VMS to the Itanium platform).
Susan SkonetskiPosted December 17th, 2013
This is a very elegant, secure operating system that was created by excellent
engineers based on what awesome customers wanted.
Tom MoeyersonsPosted December 17th, 2013
I completly agree on the following :
“If the hardware must go, consider keeping the software platform in place.”
I helped migrating several VAX and AlphaServers using the Charon-VAX and Charon-AXP hardware emulators from Stromasys with excellent results even performance…
OpenVMS is remarkable for many properties:
– logical names, which allow easy migration of devices, dynamic state variables which are globally accessible, and much more;
– a robust queuing system for batch, print, and other queues;
– an API which is rigidly enforced: backwards compatibility is built in, so non-privileged software which ran on VMS 1.0 still runs on the current 8.4;
– clustering which is rock solid and can span very large distances;
These are only a few of the great features of VMS. It is still valued by those who know how good it is. Let’s hope HP will continue the always-up tradition.
IanPosted December 17th, 2013
I have worked with VMS for nearly 30 years and in that time I have seen no end of problems with infrastructure and services, but the VMS systems just keep going, especially those configured for disaster tolerance. Data centres lose power, Operators power down in the wrong order, fibre links get dug up, Aircon dies, etc, all in a day’s work for VMS, still running, still providing the service and sometimes it can be hours before anyone notices the problem!
Come on HP, do the right thing!
Indeed. One measures an operating system platform by how rarely intervention is needed.
OpenVMS systems are stable, flexible, and reliable. The mantra of “cloud” computing with capacity driven by demand and flexible provisioning may be relatively new, but properly configured OpenVMS systems were able to provide much of the functionality decades ago and continue to do so today.
For reliability and security, OpenVMS has long been the Gold standard, and remains a solid reliable platform.
As there are no thumbs-up symbols to indicate agreement with comments, I just want to say that I agree with Ian, Parwez, and Kerry. As well as with Duane.
I like telling folks who have never heard of the operating system that one of the things that’s really great is we don’t have to worry about Patch Tuesday. They find it hard to believe.
Just because the o/s is over 35 years old is no reason to throw it out and migrate to a newer o/s. The grass is not always greener on the other side of the fence.
I totally agree with all the point of the article. It’s not so usual read such kind of articles about VMS. Thanks!
Dave BaxterPosted December 18th, 2013
I agree with all of the above. It could be another 30-40 years before another OS surfaces with the reliability, security, stability and longevity of OpenVMS. Unfortunately, we live in a world of “shiny objects”. For the last 20 years, IT has been taken over by a generation of administrators brought up on Unix, then Windows, and now Linux, and in general, their knowledge and/or appreciation of legacy systems is low.
These are the people who are now making the recommendations on IT futures, to Senior management who know even less, but who are also hooked on the “shiny objects”. (It is characterized by the “…gotta have the latest I-phone syndrome.)
IT training establishments have followed the business trend, turning out 1000′s of “shiny object” administrators to meet the business need, while legacy system administrators grow old and die. (anyone been to the boot camp lately?? See anyone under 50?)
Reminds me of an old Navy joke. “..when I joined the Service,we didn’t have Official Numbers, we all knew each other!”. The OpenVMS Administrators group is getting like that., the group is shrinking so much, we pretty much all know each other.
We have many customers running applications on OpenVMS providing mission-critical functionality on a daily basis. Such systems have consistently provided exceptional reliability, availability, and serviceability.
We enable our customers to enhance their applications on OpenVMS using the tools of their choice.
Long live OpenVMS!
R AhamedPosted December 18th, 2013
VMS rules others drool!
XTCPosted December 18th, 2013
Rather than preserving a legacy system I am interested in reintroducing OpenVMS, most likely under virtualization. I would be very interested in an article on emulating legacy hardware for OpenVMS.
Sanjay MundhraPosted December 18th, 2013
Thanks this is a good sharing on the benefits on OpenVMS, especially now with all these incidents about systems being compromised by hackers, etc.
I guess the IT Community should start embracing VMS once again to protect their Business and reap the benefits of a true Enterprise Operating System.
Live Free or Die – OpenVMS – When downtime is not an OPTION.
I have been developing software on VMS since 1980 when I took delivery of an 11/780 running VAXVMS 1.5 in 256Kb memory and 54Mb UNIBUS disk space.
Nothing has ever come close to it for security, reliability, maintainabilty, backward compatibility and richness of features.
We continue to support customers for whom OpenVMS is mission-critical.
What is needed now is for HP *(or someone else?) to port it to x86_64 architecture – all the hard work has already been done for the port to Itanium. In the meantime, emulation does in our experience provide a realistic solution to hardware support issues.
Warren KahlePosted December 18th, 2013
OpenVMS has been a solid performer for me and my employers/clients for many years and I expect it will be for a long time.
Steven ThompsonPosted December 19th, 2013
My best testimony to OpenVMS is having witnessed a datacenter disaster.
And I’m not talking a about the video we’ve all seen where a demo datacenter gets the bullet or blown up.
But in real life terms.
About a year ago, we had such an event causing dozens of systems, DBA, and Middleware people to restore service on a myriad of Wintel, *nix and other systems.
The last affected system came online hours after the “disaster”.
Of course , on the OpenVMS cluster, we didn’t notice a thing.
We finished our online, batch, backup and stats processing on time that day, as we do every day.
sudhindraPosted December 19th, 2013
Just i could not forget the amazing tool Autogen.com which provide superb inputs for system tuning since 3 decades back ie. 1983 itself. Still no technology can match with that type of tool to tune system parameters. Also DEC has wonderful software MANMAN, DTR, All In one office products etc.
Lynda JonesPosted December 19th, 2013
Excellent article. Very well written. It is reassuring to know that Nemonix has the knowledge and range of solutions to keep OpenVMS customers functioning well into the future.
Norman F. RaphaelPosted December 19th, 2013
I, too, am one of those elders. Since 1969 and until I was retired, I made my living in various ways on VMS systems and VMSclusters. I applaud the way stable systems can be preserved. Unfortunately, ISV’s have steadily abandoned the platform, so if dynamic growth or need to change software is involved, VMS use just cannot be sustained. I miss not having to do anything to my VMScluster systems now that I do not have to do anything to VMScluster systems.
Galen TackettPosted December 19th, 2013
The abundance of documentation available for VMS is something that’s impressed me from the time I began programming, integrating, and administering VMS systems 30 years ago and ever since.
Starting early on and continuing over the first half of that period, I took DEC’s training for VMS developers, courses on systems internals and device drivers, system management (or administration as it’s now more commonly known), DECnet, VMSclusters, and more. This taught me from the start to read the excellent documentation that came with the system.
Even in its earlier years, VMS’s documentation was by and large professionally published, well indexed, stylistically and editorially consistent, extensive in coverage, and detailed. It didn’t come cheap, but it was worth its price. (Not that there weren’t exceptions at times in some of these areas. Nobody’s perfect.)
By contrast, let’s look at a contemporary competitor such as Unix. There the main sources of user documentation were the “man” pages. I often found these of inconsistent quality technically and editorially, as well as far less user-friendly than DEC’s manuals.
In my opinion, a great example of VMS documentation was Ruth Goldenberg’s _OpenVMS Internals and Data Structures_. Try finding something that detailed and extensive for a typical competitor system.
Paul BlaneyPosted December 19th, 2013
Hands down, the bet OS in the market for over 30 years.
I specialise in OpenVMS layered products particularly RdbVMS, now owned by Oracle.
I’ve worked with Rdb since V1.0 starting 1985. Rdb has been the most sophisticated and reliable rdbms on the market from day 1
and is still packed with features that Oracle Classic and Sqlserver even today don’t have. Since 1986, Rdb has had the capability of pinning tables and indexes in memory and for the
last 10 years, the entire database can be pinned in memory with full recoverability, assuming you have enough memory.
I don’t know about Oracle but Sqlserver is only approaching this point with Sqlserver 2014
Like VMS and unlike other DBMS’s Rdb will run for years without requiring DBA intervention, where the environment is stable. Only when the environment changes
eg transaction rate, data growth beyond a certain level, application enhancement, do businesses find they need
a DBA to review, reorganise and tune their database(s) for the next period.
Because there is so little Rdb/VMS work, when there is demand, people find there is a shortage of these skills.
Most demand these days is in the US and I often get enquiries from there. Unfortunately, recruiters don’t understand that they could get me an
E3 work visa within 2 weeks so everybody loses.
IMHO HP and Oracle acquired DEC developed products far too cheaply and made the assumption that cheap means no good. They didn’t understand that they got a bargain, which they could use to enhance their businesses.
They then compounded their error by trying to persuade DEC customers to migrate to their own less sophisticated and reliable products. Those customers, who succumbed to the hype found that they had to spend huge amounts of money to achieve an inferior result than they had before migration.
These horror stories spread around, so that others told the salesman that if they migrated,
they would consider non-HP/Oracle platforms instead, so that type of migration pressure eased
and has been replaced by announcements that there will be no support in future for new hardware.
I predict that these companies will continue their attempts to kill OpenVMS and its layered products and
they will continue to fail to do so. Whilst customers are prepared to pay for support, which they hardly use, because the products are so reliable,
the business will still be highly profitable but with declining revenues, which is what interests bean counters. They are not interested in profits, otherwise they would have worked out by now that
VMS business is highly profitable and for a little extra effort, they could make even more profits out of VMS.
From time to time, I still come across systems running Rdb V6, which predates Oracle’s purchase of Rdb 18 years ago. There is no support for this version but the excuse is “we are going to migrate soon, so there is no point in upgrading”. Since then everything
has become more reliable and stable, so I would say this is a portent for the future and in 20 years there will still be plenty of VMS/Rdb sites, whose mantra will be “It aint broke so why fix it?”
Dave WilsonPosted December 20th, 2013
” I miss not having to do anything to my VMScluster systems now that I do not have to do anything to VMScluster systems.”
I love that quote, Norman.
JPF SebastinPosted December 20th, 2013
All that have worked with OpenVMS get surprised when some functionalities added to other operating systems and databases are presented as new, when they have been provided by OpenVMS in a highly realiable way for many many years, and the others are stil fighting. How about cluster file systems and in memory databases? How about boot from SAN?
Rick LadePosted December 20th, 2013
An excellent article. I have been an OpenVMS bigot for years! Long live OpenVMS!
I, who have in the past referred to myself as “Mister TDMS” (2001-2005), and now sometimes refer to myself as “Mister Sunshine”, heartily approve of this excellent article as well as these excellent comments about OpenVMS.
I have watched “newer and better” become the management catch phrase for “cheaper and lesser quality” over my 20+ year career. IT went from high quality platforms like OpenVMS to Alt-Ctrl-Del syndrome because manufacturers could make more money selling a sub $300 PC as an $8000 blade.
Despite the catastrophic failures associated with the combination of cheap hardware, low quality operating systems and visa workers, management keeps singing that song. In large part I believe it is because so-called “independent analysts” are paid to market it.
Every retail operation on the planet should be required to something rock solid like OpenVMS instead of some low quality x86 based OS to handle and store credit card information. The TJ Max theft was neither a one off nor an oddity. Just this holiday shopping season Target had to admit the theft of 40 million customers credit card information.
Cheaper isn’t better, it is just lower quality. The human race is paying the price for that being allowed to happen. Will any upper management from retail giants with low quality systems go to prison for the massive credit card breaches which happen on their watch? No. Will lives be ruined as people try to dig out from the identity theft? Yes. Some will be lucky enough to have it caught early after just a few charges. Others, regrettably, will suffer full blown identity theft.
That has to change.
David BeornPosted December 28th, 2013
Virtualization is the way to go with OpenVMS – then it doesn’t matter what your “hardware” is – it can run on basically any real hardware. This is one “new” concept that OpenVMS didn’t invent but it goes along with it’s philosophy of virtualizing everything. And systems being able to run on various hardware and look the same or restore a backup from one type of disks/controller to another type seamlessly. Great engineering!!
File versioning is another sorely missed feature of OpenVMS…..
OpenVMS as a “legacy” platform is only a few years older, maybe even younger, than Unix. What has worked, continues to work and work well with years, if not decades, of continuous reliable up time and availability that is unmatched by any “modern” platform.
Neither Compaq nor HP have had a true appreciation of what they inherited when they bought out Digital. The Alpha architecture, the first commercially available 64 bit RISC chip that ran OpenVMS, Digital Unix and Windows natively and later Linux, was a true innovation that met a premature demise before it flourish like it’s predecessor the VAX.
The article eloquently brings out key features, reliability, stability and security and equally elaborates that although hardware manufacturing has discontinued, the existing hardware itself has life until it has been outgrown, which in many cases spans decades.
This is a platform that lets management sleep soundly at night while consistent outages and breaches lead to marathon conference incident calls that run through the wee hours. That is the true legacy VMS will leave which went under appreciated.
Gerhard SchwartzPosted January 2nd, 2014
Appreciate the comments of those many OpenVMS users reporting here about their good experience with the stability and reliability of that IT environment.
However, it’s a fact of life that there are more so-called legacy IT environments around that do offer better-than-Unix reliability, stability and security (and are far more stable, reliable and secure than the currently prevailing “good enough” Windows/Linux stuff).
HP have the challenge that apart from said mainstream stuff they also do have the more reliable Unix implementation HP-UX and two extremely reliable IT environments (OpenVMS and NonStop). Continuing to develop and support all these environments was not longer seen as financially viable, so HP had to take a decision.
Weighing the two extremely reliable environments against each other, it was found that the NonStop environment is even more reliable and also easier to use, in addition it is less dependent on third party products (eg. featuring its own NonStop SQL database and thus not relying on Oracle). On top of that, the NonStop business is quite solid (during the last years even showing slight increases in revenue). HP’s decision was taken accordingly.
So what can OpenVMS users do in the long run ? Of course, they can settle for those prevailing and less reliable mainstream environments, or they can also go for OpenVMS emulation on mainstream hardware platforms. In either case, they must expect less operational reliability and higher operating cost resulting thereof (downtime cost, system administration cost, etc.).
Or, they can decide to move to another extremely reliable platform. IBM will point to their zSeries mainframes, but it is worthwhile to take a good look at HP NonStop which is less complicated and costly. NonStop is also here to stay, the world’s payment systems do largely depend on it …