Matt Perdue: Making things work for a long time to come
May 26, 2006
OpenMPE Board member Matt Purdue is the subject of our May Q&A interview. He pursued the Series 987 system which had a starting bid of $7 and won the system at auction. He's also put a lot of time and energy into helping HP 3000 sites homestead, or prepare for their transition.
Yesterday we ran the first part of our Q&A with Purdue. Today we continue with his reports on migration pace, optimal platform, and how long an HP 3000 could be expected to do useful service.
How fast do you see your clients and contacts doing their migrations?
At a very cautious pace. One has to remember that these application systems are used to run the entire business (at least the ones I work on) and not just a portion of the business. PCs are used, but the core business data resides on the 3000. Therefore they are very cautious in changing the system that supports their very existence.
Of all of the platforms and OSes you’ve seen, which one looks like a good fit for the 3000 customer who’s got to migrate?
Linux. Running on anything.
Being realistic, how long would you recommend a customer could rely on their 3000? What are the factors you’d ask them to consider?
For hardware and software, at least another 10 years, maybe more. The hardware should be able to be supported with parts from the secondary market for at least 10 years — start buying now. Software is another issue; if an application is relatively stable, then 10 years at least — if the app needs changes, then the support period depends on availability of people to do the work, and I’m sure there will be people around the MPE community for at least another 5-7 years with dwindling numbers as Year 10 approaches.
Other consideration factors would include hardware availability and stability of the applications. Along with availability of people to do software support and maintenance and hardware maintenance. Consider what has happened in the almost five years since HP’s announcement — people and hardware are still available and that five years has passed very quickly. Certainly many shops have gone to other platforms or are still in the process, and overall the number of software people has remained basically stable.
I remember some saying that everyone was going to flee the 3000 and there’d be no one left to turn off the lights by 12-31-06. Well, if the past five years experience can be a guide for the next five, that forecast will simply not be true.
Considering your Amisys experience, is there a packaged application you could recommend to healthcare companies leaving the 3000?
Facets provides about 60 percent of the functionality of Amisys, so for those going to Facets there will be a lot of surround code that has to be written and tested to get back to business levels available with Amisys/3000 and Amisys Advance for HP-UX. Amisys Advance is available and probably the best course, but any Amisys customer has to take into account the end of the PA-RISC chipset announced by HP last year — so they’ll have yet another migration in the future. Migrations will become a way of data processing life. I do have a group of consultants that would like to write a replacement for Amisys, but that as you know takes time, money and a willing HMO to assist with the design and testing.
On support: do you offer it for hardware or software on 3000s? Is the third-party market meeting the needs of your clients’ support?
While I maintain my own systems as much as possible (I’ve been “homesteading” since my first 3000 Series 48), I do what you could call “front line” hardware support for clients — taking a look at the problem and making a determination based on experience if the problem can be solved internally or either HP or third party support has to be called.
For example, I can replace disk drives, power supply boards, interface boards, etc. and do reloads or reinstalls with no problem, but if a CPU board goes out, well, then it’s time to call HP! Software support I’ve done since 1977, including acting as a standby systems manager from 1977 to 1985 for one company. Not to knock HP’s hardware support people, but for a long time now most of them are based in the 9000 and don’t really know the 3000. So some of the third party support companies can provide very good maintenance services, since they have a lot of the 3000 people who left HP!
What is the one HP 3000 end-of-life policy that you want HP to reconsider soonest? What’s most in need of reversal?
Allowing 9x7s to run MPE/iX 7.0 and 7.5. HP is not going to change its corporate mind and announce the reversal of its end of 3000 business, and that is that, so there’s no reason to even hope for that one.
Other issues such as un-crippling the speed of the lower end boxes to run at full power won’t be done either, due to third party software vendor rights issues and issues HP has internally. Personally, I’d like to see that un-crippling allowed on a case-by-case basis for any sites that use third party software which was not sold on a tier basis, such as MPEX and Adager. That way the interests of third party companies can be protected but those sites that have completely home-grown code and just a couple of third party utilities can extend the useful life of their machines.
HP has publicly stated they are having problems with sites testing beta patches. In a posting a day or so ago, Jeff Vance stated HP may be looking into changing their policy on allowing sites without HP software support to do beta patch testing. So there’s reason to hope HP will change and see that allowing 9x7s (the largest installed base of 3000’s) to run 7.x is in the interest of the community and HP. It would garner them a tremendous amount of goodwill in the community. Many, many sites have machines they will be willing to use to help HP with the patch testing and 7.x issues. HP only need to ask and they shall receive.
Will you stand for another term on OpenMPE when your term expires in 2008?
I’d very much like to continue to serve the community and present to HP the issues that many people report to me.
Does OpenMPE have a mission for 2006 until HP opens up its source code review process?
Yes, there are issues to resolve no matter what HP decides to do with MPE after 2008. The “pack it up, put it away” source code review project is still in process and we’re asking HP to let us help. You see, HP decided not to do some of the system improvements that customers requested; maybe those could be funded by the OpenMPE community and the work performed for HP, at HP’s direction of course. This would benefit the community, HP, and OpenMPE for the present needs of user base. It would also help OpenMPE and HP build a stronger working relationship.
HP’s source code review process is independent of the ultimate decision to release MPE, so that’s three issues: 1) Source code “pack away” review process; 2) Release MPE after 2008 to non-HP support and 3) Work on non-HP supplied and funded SIB issues. I’m sure there are others, but these three come to mind right away.
What do you think is the most misunderstood aspect of being a 3000 homesteader?
It depends on who’s doing the misunderstanding. I know of some people that think that if the management of a company won’t allocate the funds to pay for what can be very expensive migrations, then they’re not facing reality or they’re not worth supporting anyway. For those people, I’d say that I’m sure they’ve put in budget requests and made their best professional recommendations, only to be turned down by upper management for whatever reasons, sometimes despite making strong and urgent appeals to management.
There are those that think homesteaders are cheap. Well, maybe they just have different priorities for their capital than others think what should be done. It doesn’t mean they’re cheap or not worth supporting; it just means they’ve decided that what someone else suggests or wants to sell is not in the best interest of their organization.
I’d say most homesteaders I know (individuals, mid-level and upper level management) want to get the best value for the dollars spent — and spending many tens-of-thousands of dollars, or even millions, to simply move from one platform to another with almost no other benefits other than “we were on box A, now we’re on box B” is not a sound business decision.