I used to think we should watch how vendors treat the customers they have on their proprietary systems, because they would treat everyone that way if they could get away with it.
That was in the bad old days, before "open systems". Every supplier had mainframe and server systems that trapped users, preventing them from buying from other suppliers. If the user couldn't go elsewhere, then the vendor could charge them anything they could get away with. IBM, as many will probably remember, was the Big Nasty here.
Next, open systems led to a vast range of Unix systems, all of which did more or less the same thing. They ran on a vast number of hardware platforms, including processors from Motorola, National Semiconductor, and Intel, and the Alpha, PA, Power and Sparc Risc chips.
All that is clearing away now, as the market consolidates on one instruction set and a couple of operating systems. But in among all the changes, what is happening to the old proprietary systems?
Surprisingly enough, a few are still around. Not surprisingly, it is IBM that has the best survival rate. Its mainframes still hold the lion's share of business data, and are touted as Web servers and consolidation servers for hard-to-manage Intel servers. And the iSeries, which descended from the System 38 of the 1980s via the AS/400 of the 1990s, had a big facelift this week.
Meanwhile, Digital's OpenVMS -- one of the most popular operating systems from the pre-open systems days -- is in terminal decline, and unlikely to survive long after its Alpha hardware base is phased out. Most of the others have completely disappeared, along with many of their suppliers.
Why is it that the most-hated proprietary systems supplier of the 1980s is now the only one that has satisfied customers and a roadmap in 2003? The answer is in the way the different operating systems were handled.
Operating systems, like any technology, wither away. They die when they no longer have a big enough user base to support continued development, and new demands come up that can't be met.
Last year, the iSeries sold 24,100 systems worldwide, and is still IBM's largest user base, according to IDC figures. OpenVMS, by contrast, sold 6,300 systems.
"OpenVMS is older, and has had more bumps in its road," said IDC analyst Martin Hingley. With Digital passing to Compaq and then to HP, OpenVMS users have had very clear reasons to doubt their vendor's long-term support. At HP, ironically, they feel they are getting better support, even though their systems are on the road to nowhere.
Around 1992, Digital moved OpenVMS to its Alpha RISC processor, a messy and upsetting transformation. Users that moved had lots of work to do, while those that didn't were stranded with big investments that were increasingly worth less.
Then, after years of uncertainty, in June 2001 Compaq signed the deathknell for OpenVMS by announcing it would not develop the Alpha processor beyond the EV7 series. Maybe this was a political mistake. Compaq wanted to save money, but Hingley believes that it is now easier to do chip development.
IBM didn't have this problem. It had a well-established Power RISC processor in its Unix range, and the AS/400 had been built with software abstracted from hardware. The system could move across to the new processor without a hiccup.
If you like, iSeries managed the change more easily, simply because it was more disciplined in its proprietary nature. The walled garden of software on the iSeries had the effect of making it easy to move to another platform.
And within the confines of a proprietary system, it turns out that the vendor can make it comfortable for the user. ISeries users have free Linux to try out, and the option of managing NT servers. IBM can argue, pretty convincingly, that its iSeries users have a better total cost of ownership, and now have most of the benefits of a more open system.
All of which leaves me feeling unsettled. Pretty much every processor is either from Intel or uses Intel's instruction set. Operating systems are either Windows or Linux.
The open systems world doesn't look at all like the one which open systems campaigners started out in search of 15 years ago. In many ways, the iSeries has changed to survive and is as much like those open systems as other products.
I am glad to see that the once-hated proprietary systems have not yet all bitten the dust, and I'm sad to see VMS so reduced. All this is probably nostalgia -- why should users be expected to continue with something outmoded unless it gives them some real benefit? Why not get on and invent new stuff?
Nevertheless, the feeling that something is being lost is strong. The closest I can get to justifying it is that maybe it would be a good idea to keep some variety in the gene pool of computing. In which case, let's at least keep some examples in museums or computer zoos.