Cooking Up New 3000 Strategies

Birket Foster has been heating up the rare steak that is the HP 3000 community on both sides this year. The last person to win Hewlett-Packard’s HP 3000 Contributor of the Year Award at HP World, Foster has developed his company MB Foster Associates into one of four HP Platinum migration partners — at the same time he sits on the board of directors of OpenMPE Inc., dedicated to extending the life of the HP 3000 beyond HP’s planned 2006 exit.

Birket — one of those 3000 community’s members better known by his first name — has been astride the middle of the community for quite awhile, in both middleware solutions and among his peers in utility software. In the middle 1990s his company’s ODBC technology was chosen by HP to be included in the MPE/iX operating system, putting MB Foster as close to the inside of HP’s labs as any company not owned by HP can get. There’s advocacy on his menu as well. For more than a decade he’s led the SIG-Softvend Special Interest Group, a group of HP 3000 software vendors whose mission includes getting more information from HP about the 3000’s technology than the vendor would be inclined to deliver on its own.

Less than a month after HP surprised much of its 3000 community with an end of HP support announcement for the platform, MB Foster Associates started Webinars on migration issues. The briefings, which continue to take place every other week, represented the first such training anyone offered on whether to stay or go from the 3000; even HP didn’t have a briefing ready until more than a month later. By the time HP managed to air its first migration Webcast, Birket had been named to the first board of directors of OpenMPE, Inc. He’s said his objectives in joining that group include making sure HP documents the operating system code well enough so third parties can build patches after HP exits that business.

Foster’s company celebrated its 25th year of serving the 3000 community this May, starting from consulting work he did in the 1970s and then adding resales of VEsoft, DISC, and Bradmark’s software afterward, and then sales of the WRQ Reflection line. In the late 1980s he took on DataExpress from its author David Drummond, beginning the work in IMAGE data exchange that led to introducing the first ODBC product to the market in the early 1990s.

We have grown interested in the balancing act Birket and his company have unveiled over the past year of Transition. We talked with him by phone during the week that he threw his annual BBQ celebration at the company’s Chesterville, Ontario headquarters among the Canadian cornfields, hard by the South Nation River. It was a typically cascading, wide-ranging discussion about how he and his company can guide customers toward two disparate destinations, and how HP’s hope for “business as usual” is faring in a very unusual year for 3000 software suppliers.

People have been taking sides in the 3000’s Transition era. You’re on both the OpenMPE board and the Platinum migration partner list. How can you do both jobs with passion?

Have you noticed the name of our transition seminar? It’s “Should I Go Or Should I Stay?” We’re going to support people no matter which way they choose. It’s all about doing what’s right for the customer. We have a passion for looking after customers. In May we made 25 years, a pretty big milestone in this industry. We will be a trusted advisor and assist them in shaping a sustainability plan, if that’s right for the customer.

What’s the most significant goal which OpenMPE must meet this year?

OpenMPE is not about an open source code movement maintained by the community. I don’t think you have enough critical mass, those who will put their hearts and souls into it. If you could clone [Allegro Consultants’] Stan Sieler 200 times, you’d have that.

OpenMPE means to most customers that there’s a chance you could have an emulator that would allow you to run MPE on the latest Intel-type hardware for the next n years. The biggest thing that would make that possible is Hewlett-Packard not standing in the way of licensing for it. I’m sure HP has to negotiate with the 9000 division as to whether or not they could have [hardware driver] information released to the world. That information might be phenomenally interesting to someone who’s competing in the hardware business.

So you think the major goal is getting HP to give itself permission to enable an emulator?

Yes, that’s a good way of putting it. HP has to figure out that it’s okay for someone to use all this information to build something that would be able to be licensed.

What’s the goal that OpenMPE has to meet in this project?

It’s fish or cut bait time. We’ve got to get a crystal clear message from HP on the licensing issue for the emulator.

What would you say to people who believe not much has happened for the homesteaders from OpenMPE so far?

Quite a bit has happened. HP’s gone from “We don’t think that’s a good idea,” to “You know, maybe there are reasons for making this happen.” HP has more data, and I don’t think they had any data when they made the decision about the [end of support] deadline. I don’t think it’s bad to have a deadline to make people think about a situation.

Do you think that HP should be writing this hardware emulator for the 3000 homesteaders?

I don’t think they have someone with the passion that [Allegro’s] Gavin Scott has. I think you need somebody with passion to make this happen, and I think Gavin will be just fine, thank you. I think the issue has become what can we get done in terms of licensing.

The issue that the emulated mode would be is that the entire environment is not the operating system. The environment goes well beyond the operating system. It would require that there was a clear path for people to able to license and protect their intellectual property on that platform. Therefore, things like the HPSUSAN number and the HPCPUNAME would have to be maintained by whoever’s writing the operating system. Because, quite frankly, it costs millions of dollars a year to run a lab. People who are doing that kind of thing need to get a return on their investment — unless you’re doing it as a hobby.

How much of the customer base is spending money on their migration today?

We were the first to have Webinars, so we have a lot of data, know a lot about what’s going on. We’ve talked to lots of people. On November 14 there were already people who were migrating, so all that this HP decision did was confirm what these customers were doing. Those people already had a plan.

The key to this is that you must have a plan before you start down this road. You’ve got to be thinking about life after migration. This is not about migration. If you’re one of the people who wants to migrate, you’ve got to stretch your imagination about the technology that might be available in 10 years.

It’s not about the application itself, it’s the way you deploy it. You could be doing something about wireless, you know? Think about it: browsers didn’t exist 10 years ago.

Are you saying these are places where HP 3000 owners could be spending money, today, and get a return on the investment in their platform?

Yes, they could innovate before they migrate. They could get their heads around a new technology. People who own code know in their heart of hearts they should be doing something about Webification, like for order status. You’ll learn about what it takes to integrate Web services into applications. We’re helping people build data marts, for example.

Back to the question: How much of the 3000 community is actively migrating today?

So, you’ve got that group of people thinking about migrations because they were before November 14; you’ve got a group of people who are wondering if HP will do it all for them. The answer to that one is absolutely not. You’re going to have to plan what you’re doing with your IT yourself, unless you’re a huge customer and can pay HP to do it.

I would say right now there are somewhere in the neighborhood of 10 to 15 percent who were already migrating, and they’re spending money on active migration activities. The other group of customers is starting to realize the list of things that has to happen between now and when they’re successfully migrated. Those people are starting to think, “Maybe I ought to have a plan.”

Probably 60 percent of the customers have not got any plan, and not got any budget in place. What they need to be doing this fall is make sure there’s a chunk of money to get assistance to put together the plan for making the migration path. There’s a quarter of the people who are partly planned, and most of them have not thought about life after migration.

Are they spending money on that planning, then?

Some of them are trying to build it themselves. One of our partners came up with a great saying about this: “A fool with a tool is still a fool.” I don’t mean to denigrate the 3000 community. There are a bunch of bright guys. But we’re talking about business planning here. The technology decisions will follow. There are people in the market with migration experience. We just happen to have a team that’s had a phenomenal amount of migration experience.

Companies are beginning to talk about their migration experience when looking for engagements. How do you define a year’s worth of experience with migration?

Early in my career I was part of a team that had to migrate about 4,000 programs between a Burroughs 4800 and an HP 3000, and they were all COBOL. And you just can’t do editor and global change stuff fast enough. We ended up writing a parser.

The 12 people on our migration SWAT team have 118 years of experience. Some of them have done Unix to Unix migrations, some of them have done migrations to and from HP 3000s, and some of them have helped customers move and implement to PeopleSoft and Oracle.

How do you define a year’s worth of experience? Doing it and nothing else in that year?

Being in the trenches, making it happen. Most of these migration projects are 18 months, and only about 15 percent are your calls back to work on existing code.

Of the 12 people in your migration leadership team, are they doing anything else other than migrations?

Right now, yes. They are helping people innovate before they migrate, and some of them are in our tech support and R&D.

So it’s good to hear there’s something going on besides migration, and planning for it?

Oh yeah. Just because somebody is trying to herd you in a particular direction doesn’t mean that’s the right thing for your company to do at this particular time. People need to take a deep breath and assess what they need to be doing. The key thing is to do the right thing for your company — which may or may not be a migration.

Is there a good reason to buy another, newer HP 3000 before HP stops selling them?

Absolutely. The new PCI bus is getting between two and 20 times faster performance than the old boxes. You could buy an A-class box if you’re on the low-end, and you’ll be absolutely astounded if your batch jobs have been chugging away.

If you’re planning on staying awhile, the PCI stuff is going to be much more supportable 10 years from now than the previous generations using HP-IB. People don’t make a lot of HP-IB stuff anymore.

So it doesn’t seem unusual to you that there would be active sales of a system the vendor has announced an end of support date for?

I disagree. You’re in a situation where most people write their machines off after five years. So what’s wrong with doing a five-year investment? If the box is working for you with the current applications, or if you’re thinking of doing migration activities on the same box you’re running for production, you’re going to need a little more headroom. Take a look at your business plan and see if you’re planning on merging other companies into yours. If you are, how much more headroom will you need for that? You want to understand how your equipment will support you over the five to 10-year timeframe. It’s going to take 18 months to five years to do a migration if you’re planning a migration.

You hosted some tests of HP Eloquence. Is Eloquence ready to take on IMAGE duty for the average HP 3000 shop?

We had people from HP — Ross Macdonald and Tien-You Chen — and [Eloquence creator] Michael Marxmeier, and several people from the 3000 vendor community. HP first gave a blessing to Eloquence 12 years ago, when they were phasing out the HP 250s and 260s, and bought the product from Marxmeier.

Does that 12 years make Eloquence ready to take on IMAGE duty for the average 3000 shop?

Eloquence is not on just one platform. Eloquence has over 2,000 licenses, and probably only 300 of them are related to HP 3000 migrations. The layer that is used to intercept the TurboIMAGE calls is getting the best workout it’s ever had. A large number of vendors are making comments and making things happen.

It’s getting that workout just in the last six months or so?

That’s right. This is not one of those things where it was 100 percent when it was first suggested as a solution. One of the largest things about this solution is there is no HP investment in it. HP owns the intellectual property, but all of the work is being done by Marxmeier Software AG, as opposed to an HP lab being responsible. I don’t think that needs to change.

Will the 7.0 release of Eloquence change things significantly for the 3000 customer looking at the database as a migration target?

It adds quite a few things that were more 3000-ready than previous releases. Michael’s done a good job.

Will the year that Eloquence requires to mature make a difference?

Most migrations won’t start until 2004 anyway, because people will need the next year to go and plan. In 75 percent of the cases, you’re going to do an application buy, or re-buying your existing application for another platform.

Do you think HP 3000 IT staff should be getting database design experience and training?

Yes, if you’re going to develop. I think more customers will buy than develop. Unfortunately, many of the companies running HP 3000s have focused their staffing on the NT and Unix space to get the support they required. They have de-skilled the 3000 side, so they’re not up to speed on things like Apache, Java and C++.

What’s the most critical part of a transition: training in new OS, data conversion, or language skills?

Training, but not in the new operating system. One of the costs you’ve got to think about is that you’ve got to train every one of your end-users on a new application if you move applications.

The new operating system has hundreds of hours available, either self-paced or on the Internet, for any one of the new operating systems you might pick. That’s the easy part. It affects a very limited number of people, your IT staff. It’s a good idea to start your IT staff on courses now if you’re going to a new operating system.

Which of the available alternatives to the 3000 is a good idea to start learning about today?

The first thing you do is to figure out what you need from the application to support your business. If the application that best suits your business is on operating system A, that’s the one you should start focusing on.

To find out about available applications, if you belong to an industry association, go ask people in similar business what they are running. Work with your accountants. Accountants love those software selection processes. Or pick one of the Class A consulting firms, or a Platinum partner. All of us know something about software selection. I wrote my first paper in 1985 on it.

How do you handicap NT versus Unix to replace 3000 environments?

Things are definitely maturing in the Linux and NT marketplaces. If you’re going to do a plan, it’s going to be two years from now before you actually get there. You can assume it doesn’t really matter which platform you’re going to be on. Today, Unix platforms are more robust. Microsoft is positioning itself as being a major player in the wireless space, as well as integrating all of the Office products as delivery mechanisms into many applications, including Great Plains. They spent $2.5 billion buying Great Plains Software, a small and medium business accounting package. They’re doing things to set up business rules in it, so if something occurs, the following people get mailed.

What advantages do the 3000 customers get in choosing an alternative platform which HP offers? There’s a lot of Unix vendors, and even more Linux capable systems.

HP is putting together financing programs, and I know they’ve been talking about doing a financing program that if you’re working with a Platinum partner, to take the fees for the planning and implementation, plus the cost of hardware and software, and work it out over three or five years. They’re talking about it.

Is it realistic to believe a well-managed company could use HP 3000s for another 10 years? How could they do this in good faith?

Yes. The key thing being the application support over that period of time. Will they be able to support the application, and make the changes required? Ten years is an extremely long time. Technology will have changed big time. There will be a lot more wireless appliances. Object libraries will be different.

The 3000 will have the Java things, the GNU compilers. Both of those are going to evolve independent of the platform. As long as the layer is still there to support Java and GNU, and I’m sure it will be, I don’t think there’s going to be an issue. You’d be able to take advantage of everything in the Java-GNU-C++ space that exists.

However, if a new major technology came along — like 10 years ago, were you thinking about wireless? — will there be a technology that shows up that cannot be integrated back into the 3000. Over the 10 year period, the answer is “probably.”

Over 10 years it will be excellent at doing what the 3000 does: terminal-based, client-server, browser-based. If there’s a new technology that comes up, this is where OpenMPE comes in. the information on how to write the drivers and how to deliver them would be available, so you could actually incorporate those back in.

One of the things I’ve been concerned with is that as HP puts away MPE into maintenance mode, to make sure they document things. So if somebody needs to come in and do a patch, all the information is there that’s required. And you know what? The most senior guy who’s been doing patches for 20 years took a retirement package at HP. It’s alright, he’s got a pretty good understudy. But it’s not the same guy who did it for 20 years.

HP wanted us to believe after the November announcement it would be business as usual. You lead 3000 tool providers in the SIG-Softvend group. Has it been business as usual for these vendors?

It’s been a very unusual year. Not every year that has the Twin Towers experience. That has impacted a lot of things, along with the fact that we’re in the bottom of a 10-year economy. 81-82, 91-92, looks like the years we’re having these downturns. It’s good, because it weeds out some of the less experienced companies.

There are a lot of customers who have realized this year that they don’t have to move tomorrow morning. For most people, they didn’t have a budget for 2002 for planning. To do planning properly, in some detail, will cost you time and effort.

Business as usual depends on the customers. A group of customers stopped for awhile buying products. Then they realized they had to get on with business anyway. We have been doing work in the areas of data marting, and adding Web front ends to current applications. On the other hand, I know people who looked at the situation and said, “I’m not investing in my 3000 through the Transition process.” More of them than last year, and some of them are foolish. They should be looking at whether they get a return on that investment during the period of time they have left on the 3000. There’s still going to be customers on the 3000 10 years from now. You should be looking at how you can use this innovation. There’s some very smart people who wrote stuff for the 3000, and makes it easier for customers.

I know a customer who’s putting in MANMAN for the 3000 now. They decided that MANMAN, for the next 5-6 years, would be just right for this plant.

Do you want HP to release a version 8.0 of MPE/iX?

I think they should have a parking release, the one that they will be supporting through 2006. As a software vendor, I don’t want to invest in supporting six versions of MPE/iX. We’d be much happier having one version to support, if that version is as stable as previous releases. It would be in the customers’ best interest to move up to a parking release and look for support on that.

Is a 3000 emulator on Intel a good business proposition for this market? Would you test your products against such a thing if anybody could build one?

We’d take advice from our customers. If a customer said they would order a product on the emulator, we could make a business case for it. If the HP 3000 with the application you currently have on it is absolutely the right thing for your business for the next 10 years, because you’ll be able to get low cost CPUs under it. The issue isn’t low-cost CPUs, though. It’s supportability and extendibility. But the alternative is to run it on what they currently have, for 10 years. If I get beyond the 10-year point, there could be enough changes in technology that being on an Itanium processor could be an advantage. But I don’t see a huge advantage for most customers of running an emulator in the five year time frame at all.

It really comes down to what the community wants to do. As people think through this process, we’re finding less and less people who can justify a sustainability plan for using an emulator or staying on the 3000 beyond the 10-year-mark.

Why 10 years, do you think?

Things change a lot. I’ve sat in meetings where I’ve said, “And what’s going to happen in the 10-year timeframe?” People say, “I don’t care, I’m retiring by then. I’m retiring on the 3000, and I don’t care. I just don’t want to have to go through another learning curve.”

What does the 3000 community have that it can teach the rest of the IT industry?

The biggest thing about the 3000 community is that it is a community. There was always that feeling. People know each other by first names. If I say “Alfredo,” 98 percent of the 3000 community knows who that is. If I say “Bob Green,” people know who that is. That’s one of the areas where the rest of the IT industry could learn.

Can that kind of community be built without face to face meetings?

We’re going to have to learn, but the technology is evolving. We could be having this meeting with both of us on minicams. I remember much slower modems than I want to remember. But it’s hard to go for a beer in a virtual world. SIG-BAR, which is the Special Interest Group for Brainstorming and Advanced Research, lived for many years and assisted in solving the problems of the world. There’s something about discussing things over the beverage of your choice.

One of the things that should come out of this as a learning experience is that all of the pieces you need to do are the same as far as platforms. It all has to do with the applications. The HP community doesn’t have a single platform destination.

Most of the people in the 3000 community are lucky we got more than we bargained for with the HP 3000. We have a phenomenal environment with a low cost of ownership that would sit and run by itself. So, many people are in a state where they have the attendants flying the plane. That’s going to cause them a problem, because we have to change direction.