January 2004

The Work to Master Something New

Alan Yeo displays a canny sense of timing. As the founder of the software companies Affirm and ScreenJet, Ltd., Yeo has contributed to the HP 3000 scene since 1980. But his work in the last several years has brought him into sharper focus for a community now separating over the fault line of migration and homesteading. Most recently, Yeo organized the World Wide Wake, a collection of places where the HP 3000 faithful could gather on Oct. 31 to commemorate the system’s last HP-related milestone, the end of HP’s sales. Earlier in the year, Yeo asked out loud where else the HP 3000 community might gather in a user conference — a question he posed in a meeting at the Atlanta HP World, where few 3000 customers had appeared. He also shows a keen wit in his communications, engaging a cartoon artist to create trenchant commentary in his company’s ads — like comparing HP CEO Carly Fiorina to Cruella DeVille from “101 Dalmations.”

Yeo also picked up the pieces from the effort to market ScreenJet, developed as a connectivity product and sold by Millware.com until that marketing company went bust during the dot-com implosion. ScreenJet remains the most prominent product from Yeo’s developers at Affirm and ScreenJet Ltd. The software earned a recent award for migration solutions from Acucorp. But for all of his effort toward helping migration customers, Yeo remains pragmatic about a 3000’s transition possibilities. ScreenJet achieved its best technical release just one month before HP announced its withdrawal from the 3000 market, and the product’s development to that point was not driven by any need to move companies away from the platform.

Now Yeo is taking a role as producer in a new feature for 3000 customers who’ve been long abandoned by HP: Transact users. The advanced development language was kicked to HP’s curb in the middle 1990s, but sites continue to run extensive Transact applications, long after the “strategic” badge fell off the language. A new initiative from Yeo and a pair of experienced developers would give Transact sites service and tools to move programs to COBOL, a way to prepare for the journey away from the HP 3000.

Yeo’s candor, timing and wit make him an asset to a 3000 community which is still listening for frank advice about its options for the future. We wanted to ask him about the prospects for migration tools, what MPE tool providers like his company think about sticking with the marketplace, and how the market divides out among migrators and homesteaders. We spoke in the days before Christmas, with the announcement of the Transact to COBOL product just in the offing.

Why did you organize the World Wide Wake around the 3000’s end of sales, and what did it accomplish for you?

It was an event that ought to have been marked, because it was significant. Up until that date, it was possible that we all could have been dreaming, and someone from HP would say ‘We changed our minds.’ The end of sales was a point in time, and I don’t think there’s going to be another point that you can mark. From now on, we’re into shifting sands where various things go off support at different times. HP’s end of support in 2006 is going to be irrelevant. People will be migrating or homesteading and will be set up to survive without HP.

Why do you think the 3000 customers have moved away from the platform so slowly, so far?

The real question would have to be why do you think the customers would have moved any quicker? The solutions are only now becoming available and good enough for anything other than a serviced migration. I also think that it is starting to dawn on a lot of people that they aren’t thick, and that given some tools to help, they are as capable of creating and maintaining solutions on other platforms as they were on the HP 3000. The people already working on different operating systems and with different languages aren’t any cleverer than HP 3000 developers. How long did it take before they became productive working with the HP 3000 when you first started? I’ll bet not very long. The same applies when you migrate, initial high learning curve but then before long the curve flattens out and your cruising again. It can even be fun!

Also, I think a lot of people from the HP 3000 community have a lot to teach the wider community about building good robust business applications and environments.

If a customer cannot find a replacement package for their application, when should they consider migration?

If you can find a real good replacement package, then you’d probably be stupid to migrate. If you can’t find one, then I don’t think anybody outside your company can advise you the right time to migrate.

About the only thing I really agree with HP on is that they should do as near as a one-to-one migration as possible, and that is probably the quickest, least costly, least risky solution. But that is only true provided that you end up with a solution that will provide you with a future development path, and you’re not going to end up locked into a technology that is even more proprietary than the HP 3000’s.

If you think the services and tools are expensive to migrate relatively common HP 3000 software, just imagine what it will cost to migrate from some completely proprietary migration solution again in a few years time. New package implementation and application re-writes really are the most risky and costly options.

As an MPE tool provider, do you want to continue serving the 3000 market? HP has some ideas that not even the utility companies want to carry on.

The question is whether it will be economic for companies to continue. Once you’re in a homestead mode, nothing dramatic is going to change. If your environment is working now — providing you can keep the hardware running — is it an environment you can keep running indefinitely? Will those people carry on paying for software support? I don’t know about the US, but certainly in the UK I know a lot of sites that have whittled their support right away in the last two years, down to the things they really need.

But do you want to stay in the business of providing technical support for people who use the MPE environment?

The answer for us is because we’re small, if it still pays, we’ll do it. The biggest risk is going to be the disappearing syndrome. If you’ve got a piece of software on your 3000 and you haven’t been paying support on it — and in three years you decide to put it onto an N-Class you buy on eBay or from a broker — you may find this piece of software that’s been playing a pivotal role in your organization doesn’t work, and the people who wrote it aren’t there anymore.

I hope that if companies decide to exit from the 3000 software market, they would sell or transfer their products’ support to companies who will be in for the long haul. Given how cheap most of us homestead sites have been, that community is unlikely to be spending to keep much of the technical infrastructure alive. There are very few people in our community who can be altruistic, and will help support people for nothing. There are a lot of us around who will help even when economically you wouldn’t do it. Mainly because of the people.

Given we’re in the migration business, we have more of an incentive to stay around and support the community than people who already have other markets. Every homesteader is a potential customer for us in the future. We hope to be here for the long haul, and we’ll be involved with the 3000 community as long as there is one.

What was wrong with the 3000 marketplace model from an economic view? Can it be fixed?

Most software on the HP 3000 was too expensive, compared with other platforms. However, because people could reliably write applications for the system, many of these were developed far too cheaply. Many customers got far too much for the money they actually spent.

Is there a need for another market and user event to draw HP 3000 customers, now that HP World has drawn so few in 2003?

Some people are going to be homesteading for an indefinite future. Some are going to be medium-term homesteaders, who know they’re going to have to do something. There’s another group that have actively started their migration, or working out how they will. What your issues are will depend on which of those phases you are in. There isn’t going to be anything new for the 3000 significantly, so you’re not going to go to HP World to find out anything for homesteading.

The other two groups are going to come up with the same questions year after year, depending on where they are in the process. HP World is not really a forum where you can repeat the same stuff year after year. The sessions at the last HP World were stunningly good. But they don’t want the same sessions the next year.

HP World has got bigger and better things to do. It’s not an HP 3000 conference anymore. I think something like a new HP 3000 user group is needed. OpenMPE could be it, but the group’s title is too restrictive. Something is needed to allow the retention of both information for homesteading, plus the dissemination of information about migration.

How do you divide up the 3000 market, and how big do you think the market is today?

The market feels like there are a third of people who don’t even know HP has discontinued the platform, or that decision won’t alter their plans at all. They’ll carry on using it for as long as they can, because that course makes the best economic sense.

As to the size of the overall migration market, I would be very surprised if more than 5,000-10,000 companies world-wide have applications that should be migrated. For the others gradual replacement, function by function, application by application, will occur.

There are another third who were already thinking of migrating or implementing some global solution. They were almost in migration mode before the announcement. Then we’ve got a middle third of companies that really do have HP 3000 environments. In my experience, most companies have built an environment around which their business operates, a combination of either applications they’ve bought and then developed and bespoke applications. They’ve got some fairly unique value in what they’ve got, and these companies looked pre-Y2K, and couldn’t find anything to replace what they have.

But there aren’t any new applications anymore for the HP 3000. At least not the global, ERP kind of applications that are running on the systems today.

Are the tools ready for a complex migration, one where 20-30 apps and spool-print subsystems have to be migrated?

I don’t think there will ever be tools in place that will pick up a complex environment and allow you to press a few buttons and presto, you’re someplace else. It like saying if I took all these engine parts out of Ford engine, which General Motors block can I stick them in and expect them all to work like they did before. It’s never going to be that easy.

There are lots of people who have written their own print spoolers that do incredibly complex things. If you were starting on another platform today, you wouldn’t do it that way. But the problem is that once you’re really making use of something like that, and then you’re working at making off the shelf software from Windows, Unix or Linux do the job, it’s not easy.

I think migration is almost like a compromise. You can take it piece-by-piece and re-engineer it, replacing those bits you have to throw away. Sites that have something like Security 3000 and MPEX to manage a lot of things in their environment — those tools don’t exist anywhere else. To migrate your environment, you’re going to have re-engineer all that.

People have had applications migrated, up to now. But if you look at any of those projects that are out there quoted as major migration projects, most of them were started long before HP announced it was discontinuing the HP 3000.

What changes have you seen in the level of interest in migrations over the last quarter?

In the last quarter of 2003 we have seen a dramatic increase in people doing migration studies and evaluations. All of the migration service partners are now working on projects where the customers had declared before they were going for replacements or re-writes. It’s because the inexpensive part of replacement is buying the package. The business change involved is the enormous cost. If you can move what’s of value to a new environment, so it operates as it did before, it’s probably going to be a lot cheaper, smoother, and certainly a lot less risky than the other options. If you make sure where you end up is some place you can move forward from — to do the things you couldn’t with your 3000 — you can end up better off, rather than worse off.

Your ads talk about being “a master of one.” What does this represent for your company, and why is it a better guideline when choosing a solution supplier?

We’re starting to see more collaboration between migration tools providers and migration service partners. To get some of this stuff right, you really, really need to know it. I think it’s too big for any one person to do anything right. If you want good fish you go to a fishmonger. If you want good meat, go to a butcher. If you just want food, go to Wal-Mart, and if you just want to eat, you go to McDonalds.