Since last month the problem of older VS systems has been nagging at me. This is the problem of user organizations outgrowing a VS that is not upgraded. It creates what I think of as "The Self-Fulfilling Migration Illusion," which is the perceived need to move to a new and different platform merely because the old one has been allowed to age to the edge of usability. It's a little bit like convincing yourself that you have to buy a new house because the old one isn't as hospitable as it used to be, now that the furnace and A/C and roof and dishwasher are getting old.
If computers were houses, we would readily upgrade the A/C to get better cooling at lower operating cost, and we would understand that roofs and dishwashers and the like have to be replaced periodically, sometimes bringing welcome improvements in performance. So why is it that we so easily go brain dead when it comes to Wang VS computers?
Before I go into that, let me get one thing clearly on the table: The VS is the most scalable mainframe in the world. From the smallest model to the largest, the VS provides mainframe computing in a friendly, easy package – so friendly and so easy that it may also be one of the most underrated computer lines in the world.
A Mainframe in Disguise
The VS shares a common heritage with the workhorse IBM mainframe models, the 360/370 and, probably, the present S/390. Its machine and powerful macroassembler languages are instantly recognized by mainframe programmers, as are its mechanisms of print and job classes and background task initiators. It has a sophisticated job control language and an integrated program development environment that supports nine or so programming languages. All models run the same object code. A program developed on a tiny system under someone's desk can be loaded and run on the largest system and be used by many hundreds of users.
VS systems can be clustered and share disks, files and databases and can route print files and jobs to each other by class. VS networks supporting seamless remote logon, file transfer, email routing and application-to-application data pipes can and have achieved sizes of 900+ systems with tens of thousands of users, fully synchronized directories and adaptive routing of traffic around congested or downed links.
The VS is a mainframe. You can get it in a big, fast model or a small, slow model. You can use the features you need and leave the ones you don't. I think it is fair to say that few customers have fully utilized the features of the VS, and most customers have utilized only a very few.
Why the Rush to Migrate?
So how is it that the VS has become a target for retirement in many organizations? Aside from the general pressure on large computers that I discussed last month, aside from the nervousness engendered by Wang's 1992 bankruptcy, and aside from the OS unbundling and pricing issue that alienated many customers, there is something else: It is the tendency of a segment of the VS user community to take the path of least resistance and just do nothing to keep its VS systems current.
Consider that Wang never mastered the IBM style of aggressive, customer-visiting, CEO-cultivating salesmanship, except possibly with its largest customers of the 1980's. Consider that Wang never planned its customers' upgrades as part of its back room marketing strategy and then followed up by hiking existing maintenance prices and pressuring customer management to buy the new models, the way IBM did. Add the aggravating factors mentioned earlier and an blizzard of unbundled, extra-cost software options, and the stage is set for customer indecision. Some of those who were able to avoid or postpone dealing with it, did nothing, and the deep-rooted strength of the VS let them run the systems way past the point where upgrade would have been the right technical move. Some others actively refrained from keeping their VS systems up to date because they were unhappy with Wang over one thing or another.
As with disks and their operating costs, where it may cost more to continue running older disks than to replace them with newer ones, so it goes with entire VS systems in not a few cases. The tragedy is that, as in my example of the house, for lack of reasonable attention to keeping the VS current, all too often it becomes seemingly justifiable to trash it and buy a new and different system. Sometimes that works out, but all too often it is an irreversible course of action that holds many unpleasant surprises and penalties. It is doubly tragic when it happens while extraordinary VS bargains are available, in excess of the predictable performance / price improvements naturally occurring by Moore's Law.
If someone were to propose spending $10 million or more to migrate from a computer when the largest model of that computer in the world is priced in the lower half of six figures, they would be shown to the rubber room, right?. Yet more than a few customers have ended up spending that much or more to replace a VS, sometimes having to live with a combination of hardware and software decidedly inferior to what they had when they were on the VS. It's especially bizarre when it's driven by someone in management who doesn't "like" Wang -- someone who may not even stick around to catch the heat from a poorly-conceived migration (but that's yet another story).
Unraveling the Opportunities
With all that in mind, I thought I'd try to make sense of the many options available to customers who have less than the latest and greatest in VS technology, and see if any clear avenues presented themselves. I did what I could to get a sense of the present market, the maintenance costs, and the bottom-line VS horsepower-per-dollar available today. Keep in mind that my numbers are relative coefficients, and are not precise. Also keep in mind that "your mileage may vary" because the value of your present OS license and the potential trade-in value of your present VS can greatly affect the numbers.
The ratio of performance to price gets better as equipment of any kind gets older and is available in secondary markets, whether from the manufacturer or from third parties. The ultimate performance for the dollar is to be found in the oldest models that no one wants, but there are reasons no one wants them – high maintenance and operating costs, too little performance to be useful at any price, and possible absence of support altogether. For this reason, I didn't consider the oldest VS models, which are also the ones ineligible for the Y2K OS. The models that are not viable are:
VS 50, 80
VS 85, 90, 100
VS 5, 15, 25, 45
VS 6, 6E, 65, 75E
It is also happily the case that replacing any of the models above with any newer model will result in greater performance, with the sole exception of going from VS85/90/100 to the very slowest members of the VS5000/6000 family.
Of the many remaining viable models, including the newest ones, current performance / price ratios range from 10 to nearly 90 times what the first VS offered. Adjusted for changes in the dollar (the only type of comparison that is ever meaningful), the real improvements are from 30 to 240 times the original VS80. As expected, older models such as the various VS5000s also offer the most dramatic increases in their own performance / price ratios since they were first introduced, owing to their present low prices in the market and their modest OS license fees.
Perspective
A few historical facts are helpful in following how the numbers work. VS systems for the most part come in large and small cabinet formats. Performance / price (P/P) at the time of introduction has been distinctly different for the two formats and sets the stage for some of the most advantageous opportunities.
Relative P/P of large models was remarkably constant until the VS12650, when it jumped up considerably, perhaps a necessity in Wang's more difficult post-1992 market. The VS16850 seems to represent another big improvement, and I'd guess that the impending VS18000, though no doubt pricey on the surface, will set a new P/P record for Wang.
The small cabinet models have always had much better P/P ratios than the large models. That has been offset by their rigid upper limits of processor options and IO expansion. Still, P/P ratios in the VS5000/6000 family seem to have been two to six times better than those of the large models, probably necessary in the small system market where customers are inherently more vendor-mobile.
The VS6230, while it has the cabinet and IO in common with the original VS6000 family, is really a different and third type of creature. It has a big-machine processor and, like the VS12650, is a post-1992 product that offers much higher performance for the price – in fact, higher than any other VS models ever released. With the Turbo option, as the VS6230T, it offers 35% more processor power than the fastest VS7000, 8000, 9000 or 10000. Both in raw power and P/P ratio, this family is in its own class, way above the original VS6000s.
Uncovering the Gemstones
What matter today, of course, are the P/P ratios based on current market prices, not the ratios as they were when the models were first introduced. Performance / maintenance ratios may be important to some customers as well -- if a model is cheap to buy, affordable to license, but costs an arm and a leg to keep under maintenance, its utility value is diminished for those with tighter expense budgets. In such cases it may be more beneficial to pay more on the front end and be assured of affordable ongoing maintenance rates.
With these and other variables and conditions in mind, and with the caveat that my input numbers may not all be on the mark, here are the gems. The numbers to the right are my calculations of constant-dollar performance / price ratios relative to the baseline of the VS80 as 1.0 (for instance, the VS6230T, 20 times faster than the VS80 by published figures, is actually 51 times more computer for the money, because it is also less expensive). Higher numbers here mean more computer for the purchasing dollar, and generally take the OS license into account.
I believe that the most advantageous VS models to presently acquire and license, for the horsepower, are:
VS5000
220-239
VS6000 (16-user)
197-215
followed in distant second place by:
VS6000 (32-128)
57-127
VS7x80 VS8x80 VS9x80
75
followed in third place by:
VS6230T
51
VS12650 (if available)
47
VS6230
43
For comparison, the current very large VS models come in at:
VS16750
30
VS16850
30
Since license fees are the largest component of the cost of an older system, the value of your existing license eligible for upgrade can make some of these ratios even more favorable.
All the numbers presented here are merely a guide – something to spark creative thinking. Your present system also greatly affects what you can do and what it will cost. If you have a VS300 and your system can live in a small cabinet, you can downgrade to a VS6230, meaning no additional license fees, and performance will improve significantly. If you have a VS12550, you may be surprised at how much value it has toward an upgrade to a VS16850. If you have any variety of 80-level processor, you will have the greatest model mobility due to the value of your software license. Clearly, too, users of the older, non-viable models that will not be supported for Y2K have some relatively painless options in the VS5000/VS6000 family, in some cases representing a smaller outlay than some people have spent on PCs for home use (a fully-loaded PC in 1989 could easily run $5,000 to $10,000, which would be $6,600 to $13,200 in today's dollars).
If base maintenance expense is your primary concern, the ranking is different. The numbers to the right are relative performance / maintenance ratios on an arbitrary baseline. Higher numbers mean more computer for the maintenance dollar.
VS6230T
18.0
VS6230
10.0
VS5x60 6120
9.0
VS5x50
8.0
VS5x30 VS5x40 VS6110
7.0
VS8x80
3.6
VS7x80
2.4
VS16750 VS16850
2.0
VS300
1.2
VS10100
1.0
Note that the performance / maintenance ratios for current, top-of-the-line VS models are not "bad" – they just reflect present-day costs for present-day, high-performance systems. What's bad in the figures above is running a VS10100, a VS300, even a VS7x80 or 8x80, if it's not necessary to have such a large box or so many high-performance IOCs.
Those large cabinets were built to accommodate lots of native VS workstations and printers, and lots of disk drives that would be considered too small by today's teenagers. Today's PC workstation emulators and printers are commonly connected en masse through LAN gateway PCs, while today's disks are five to 20 times larger than older ones and are more densely supported by fewer controllers. What often counts most in a VS today is therefore its processor, not its cabinet size. Smaller cabinets, though, mean lower maintenance costs, smaller footprint and lower power consumption, all of which give the VS a better chance of survival and of delivering worthwhile service. The one unquestionable reason for continuing to run a large-cabinet VS is the anticipation of upgrading to a VS16000 processor. That's typically a one-hour field upgrade, but it isn't an option unless you have a large cabinet.
The VS6230 family is the hands-down winner when it comes to economical operation. Its power consumption and footprint are also very agreeable, on a par with common PC LAN servers. All the costs of operation are so comfortable that it's difficult to imagine ever retiring a VS6230 for any of the common reasons.
For the ultimate in reducing operating costs, those with large-cabinet systems and SMD disks may wish to look at the possibility of moving to a small-cabinet VS with SCSI drives. Even the 5:1 or 10:1 improvement in VS operating cost is exceeded by the 400:1 to 3700:1 improvement in disk operating cost.
If you have an older VS and find yourself trapped in an endless cycle of slow response, long backup times and agonizing, overnight processing while management fidgets about high operating costs and makes noises about the need to "get rid of the VS," remember that you have a significant, mainframe-class asset that may only need a bit of creative upgrading or downsizing to become a comfortably efficient contributor to your company's bottom line.
—————
Thomas Junker is a VS software developer and consultant in Houston, Texas. He may be reached by email at
. His well- known Unofficial VS Information Center website is at www.tjunker.com/.