Duane Percox got his 3000 higher education around applications, specifically those for the education administration market. The founding partner of software vendor Quintessential School Systems, Percox has been crafting 3000 applications, and companies which offer them, since the early 1980s, when he was creating K-12 apps for Pertaine Systems in California. He started his 3000 career porting a COBOL payroll application to the platform in 1979. The system got popular with school agencies in California, and he became a consultant and then part owner of Carter-Pertaine Inc. In 1990 Percox launched QSS, a new company founded to service Carter-Pertain customers in the Western US. In the last year he’s become a member of the MPE Forum, an executive steering committee of Interex devoted to improving the system’s prospects in training and development.

Percox has a good resume of both business and technical accomplishments in the 3000 market, experience common to successful long-term suppliers for the system. Besides leading a reseller authorized to sell new e3000s, Percox also has an affinity for software architecture. This spring at the SIG3000 roundtable meetings, he wondered aloud about the fundamental value proposition of the software tied to the 3000 — and how it might be updated to keep the system in step with offerings from Unix and Windows, places where applications are often created these days. In the last year his firm also closed a deal to take on more e3000 customers from Arley Carter’s Carter-Pertaine Systems, another K-12 ISV, as a separate entity of the Quintessential School Systems business operations. Percox’s firm extended the QSS investment in the e3000, to more than 110 companies installed, at the same time that he raised questions about the development tools and began supplying help to improve the platform’s offerings. We asked him about what needs to appear to revitalize the 3000 as an applications platform, and where he sees the market headed in the years to come.

QSS recently got into the 3000 market even deeper by taking on the customers of Carter-Pertaine, whose code you license in your own QSS applications. Why did you take on more e3000 responsibility, and what can we infer about the prospects of the market from your action?

We did it for a couple of reasons. One, Arley Carter wanted to retire. And just out of selfish reasons we didn’t want the company to fall into someone else’s hands, since we’re so tightly linked. Secondarily, they’ve got some good technology they’ve developed in the student information system in their ViSTA product. And they have a lot of people who know the 3000.

We’re committed to the 3000 market, and aren’t actively developing solutions that would be running off of the 3000. We could have taken that same money and bought a software company that’s got an NT, SQL Server kind of solution. But our focus has always been installed base first, new business second. That means we don’t always make HP happy, because we’re not as aggressive chasing new business as their marketing department might want. But we have a much better chance of a prospective new customer being successful when getting a reference. In fact, we rarely qualify our references. We just show our list of customers. It’s not to say we don’t have our problems, because no one’s perfect; we’re talking software here, and we’re talking people.

If you think about it, our focus on the installed base is another good reason to acquire that stock. A number of the customers he serves are customers which Pertaine Systems sold. We know those people and have talked to them since the purchase — and it’s like old home week. QSS has customers we sold in 1982. This is a relationship business.

Is it more so in the K-12 world? Do people hang on longer in their organization?

In a lot of them, the individuals have changed. But schools by definition are conservative, and change is hard and expensive. You need to be motivated for change, so we try to do our best to keep them from being motivated for change. We also have a number of customers who are not technical, and the 3000 works well with that market — because you can just turn it on and let it run.

How essential are new applications to the health of the 3000 community? If a new K-12 application provider were to enter the market from the AS/400 world, would you see it as a good thing for the platform, even if it made life complicated for you as a K-12 supplier?

Are you asking me to put on my advocate hat on one hand, and my business ownership on the other hand? It makes sense generally that the more applications you have for a platform, the more traction the platform has. It would be great to have more applications on the 3000, even if they were competing applications. That I’m not so worried about.

But I have to tell you that your example is probably not one which would occur. Most AS/400 people have much larger installed base, and a better support network from IBM as far as their developer support. I’m not sure they would be looking to the 3000 as a market play. There isn’t enough of a platform differentiation between an AS/400 and an HP 3000 for them to switch. It’s a lateral move.

So are we then talking about a Wang, a Data General, a Prime software vendor? Who’s left to join the 3000?

We just sold to somebody who was on a Prime system. It was a package we used to compete against in the 1980s, and they went out of business and a bank took them over. The bank was then running the software company, and we sold our software to the school system, and they’re going to buy an N-Class.

I think it’s been too long since those systems went under for there to be anybody to move to the 3000. If I was a vendor on those platforms, I would choose NT before I’d choose anything. It’s size of the market. The reason we still like the 3000 is that it has a breadth of performance that’s binary compatible — such that I can put a 3000 in for a small school agency, and I can put a big 3000 in for a humongous school agency, and I don’t have to worry about performance issues related to the small vs. high end. The only issue is price.

In the olden days, a low-end HP 3000 couldn’t compete against a VAX. That was solved many years ago. With IMAGE and a reasonably well-crafted application you can still get a fair number of users on that size system.

Architecture as it relates to applications has been central to the messages you’ve given at the Solutions Symposiums in the last two years. What kind of a difference can a good application architecture make to a 3000 customer?

The way you craft your application determines how easy it is to write for your application. This determines how successful you are in deploying your application. This is one of the reasons IMAGE and VPlus were so important to the 3000. These were built-in tools that came on every box. They allowed developers of not any great brainpower to craft reasonable applications without having to write all that stuff on their own. That allowed a much greater wealth of programmer-analysts to develop applications.

At SIG3000 you talked about a new value proposition for 3000, something as compelling as the VPlus and IMAGE bundled with every system in the late 1970s. Why is this important, does it need to be updated, and what would you like to see on the box?

What I was trying to get at is the feel, with IMAGE and VPlus, of using those tools to craft an application. Those applications were fine for their day. Now I want to develop a distributed application. There’s horsepower sitting all over the enterprise in Windows desktops. I want to do my screen IO in Visual Basic, but I still want all my server stuff to be resident on the 3000. It would be really nice if HP had taken a leadership role in figuring out a solution for the application developer — the ones who aren’t going to be spending all their time figuring out sockets, and who don’t have time to evaluate hundreds of different tool kits — and say, “We can make the 3000 operate in a distributed environment. Here’s a way to do distributed development that will be deployable and it will allow you to develop easily in that world.”

What was raised at SIG3000 was that the current world is way more complicated than that old world. I agree the solutions are going to be more complicated than they used to be. But relative to how complicated it is to do [distributed computing] on your own, it would still be simpler for HP to offer a solution.

Is there a need for a new common interface for the 3000? HP’s fallen back on “too many to choose’ in defending why it’s not advocating a VPlus replacement. Is its argument sound?

I understand their position on that. But if you make it easy to tie the client and the server together, then it’s a non-issue.

You could have written your own screen handler back then, but you had something called VPlus. You used it because it was on every box. As a software developer I didn’t have to worry about it, and this is a beautiful thing for a software vendor in deployment. It’s one of the reasons Microsoft is winning the desktop battle: Developers for Windows know a certain functionality will exist on every desktop. I’m saying, why can’t we have the same thing on the 3000? Why can’t I have a set of tools that lets your average COBOL developer develop applications that can work in a distributed environment — more easily than having to figure it all out on their own? In the olden days, HP took a reasonable leadership position in that. There isn’t anybody championing that in HP anymore, because HP doesn’t have a need for it. But the need still exists.

On MPE you have to be more creative in how you do the architecture of your application. It would be really cool if there was a way for me to create a server, and when it ran the operating system knew about servers, and those servers could exist in their own space, but they would be linked to a listener and wouldn’t have that job-session problem. There are ways to architect around that, but you have to know what you’re doing. It would be much better if it was a fundamental execution model of the 3000.

I also said at SIG3000 it would be great if there was a transaction bus that let you glue your server code to your client code. So that transactions could be sent from clients to servers without you having to manage that yourself. It’s not a trivial thing to do. But it has merit. Some would argue that Java is providing that.

Distributed computing has great potential for the 3000 base, from what I hear you saying. You advocated HP’s adoption of CORBA for this on the 3000 years ago, but not much has happened. What’s the upshot of this lack of embrace of an object standard for the platform?

My idea of CORBA is that it gave me the ability to host a distributed object or application that could talk to a client in an industry-standard way. I was more interested at the time in having HP consider and adopt emerging technology, to show the box was a player in the current world. It wasn’t necessarily that I had to have that functionality. I would have used it. I don’t think it’s a big deal for the average customer. That’s why I’m always a proponent for the technologies like Java to exist on the 3000, as a good retention policy. I’m not quite sure that it plays out as an attraction policy. If you have a Java Virtual Machine running on the 3000, does that mean you are going to have more people targeting the 3000 for applications?

Is there value in Java in application development on the 3000? Or is its value in another area?

Long term, or now? I don’t think it’s there yet. I believe it has potential long term, and we’re monitoring that. It’s a double-edged sword for CSY. What is today a customer retention strategy then allows the customer to evolve their application to an arena to allow their application to run on someone else’s box. If CSY does not provide boxes of the same performance as competing boxes, then what’s to stop that customer from migrating to a non-3000 solution? As you attract your customers to develop into these kinds of tool sets, you eliminate more and more of the value added of the platform.

The most highly touted Java tool these days seems to be Enhydra. What difference is Enhydra going to make to the average HP 3000 shop? HP’s Mike Yawn described it as the killer app for Java on the 3000.

The average 3000 shop isn’t going to be doing Enhydra stuff. It doesn’t mean it’s bad to have Enhydra on the 3000. Mike Yawn wants people to run Java apps on the 3000, and Enhydra is a Java app. From his perspective, it is a killer app, because it uses the Java virtual machine. But why would I run Enhydra on an e3000 when it’s publicly known that an AS/400’s Java Virtual Machine is faster? If I was just looking at platforms, why would I choose the 3000? I’d have to already have one. Enhydra is a retention strategy, not an attraction strategy.

Is there something about the 3000 still remaining which makes it a better choice than other platforms?

It depends on what you want to do. Comparing the 3000 to the AS/400 as a COBOL database application environment, there are compelling reasons for both. The issue for a software vendor would be what you know. If I were comparing the 3000 to NT, the big differentiation is that you have a box in the 3000 that knows how to play in the enterprise. While they tout NT as enterprise, it really has more of a departmental focus. If I have 2,000 to 5,000 people banging against a database server, I’m going to be a lot more comfortable with it being a 3000 than being NT.

We run a lot of NT servers at QSS. They run fine, but we are regularly rebooting them, doing things with them we don’t have to do with the 3000. It still requires more fiddling than the 3000.

HP says its A-Class is going to remain at its current speed to protect the value of the N-Class. Is there a way for you to sell an A-Class system to a new prospect in its current configuration?

We have customers that have 9x7s and are buying A500s. An A400 is 65 percent faster than a 918, and you get an unlimited user license. From my perspective, if they’re a small site and look at the dollars, it’s a no-brainer. I do agree there has to be a range of solutions in a platform. I believe it’s important to draw a distinction between the CSY business model and the 9000 business model. They sell a lot more boxes on the 9000 side, because they’re used for more different things.

It’s a different market over there. They’re competing against Sun and SGI, not against the AS/400. The N-Class equivalent boxes are in the same processor range as the e3000 models. It’s unfair to compare the two, because they’re different business models. The profit margins have to be higher at CSY for each system, because they’re not selling as many. I don’t believe they’d sell more A-Class systems if they were faster. I just think it would cannibalize their higher performing systems.

Please comment on a pipe dream: user-maintained MPE, in Open Source, to deliver enhancements for the e3000.

That’s an interesting question, because the Open Source movement in Linux and databases has been interesting to watch. As a software vendor we’re trying to keep abreast of the thinking in the industry. The concern I have as an application vendor is that if I had any Open Source tool I was depending on, I would have to add engineers whose sole responsibility was to manage our build and distribution of that tool within the customer base. If you start talking about an operating system that is as feature-rich and robust as MPE, it would be difficult for us to find that being comforting.

In other words, I’m not sure we would get the value we want out of the 3000 — it’s stable and predictable, and it continues to evolve in a reasonable way. That’s my application vendor hat. If I was a tool vendor or I was hacking around in code or something, it would be cool. But I’m not sure I want to download or ever get a distribution of Open Source MPE and try to build it, and actually have my customers rely on it — because I’m pretty familiar with my skill set as it relates to operating system development, and that would scare the bejesus out of me.

Many of the utilities for the 3000 now fall into that Open Source category. What’s to become of things like Samba, PHP, Gnu C++ and sendmail without HP’s support of them?

These are tools you can use to make your life easier for developing on the platform, not the OS itself or the fundamental database. Those kinds of things are a little different. It sounds like a business opportunity, but not for HP. Why doesn’t someone step forward?

If we’re talking stuff that’s layered, like PHP, Samba, the Gnu compiler collection, those things can be ported to the 3000 and people should be willing to support them — either as a collective effort or some business model a la Red Hat. But people are discovering it’s hard to make money in that world. There are a lot of Linux vendors whose stock is selling for under a dollar.

How has selling the 3000 to new prospects changed for your company over the past three years, from pre-Y2K to today? What are the factors you face in making new sales for the box?

I could answer the question going back five years. In a couple of isolated cases we got Y2K business. But sales is not a significant part of our business. Our revenue model is more support and service than it is sales. We find it easier to justify the 3000 now than five years ago, because of all the tools that have been ported to the 3000. It makes a world of difference to tell someone Apache runs on the 3000. That blesses the platform. It justifies the platform is not a thing that nobody’s doing any work for anymore.

The marketplace we work in is different than the shrink-wrap marketplace most people think of when they talk software. The sales cycle is very long for school agencies. We have two new customers implementing entry-level N-Class boxes, and both of them have had sales visits going back close to two years. That’s why it’s good in our business not to depend on sales, because it’s easy to have good prospective customers that just drag on in the sales cycle. You have to build your business around being in business even if you sell nothing.

How much influence do you believe an installed base of perhaps 10,000 companies have on a $50 billion company like HP?

My experience as a software vendor and someone active in the Interex user community is there is a role to play in customer advocacy as it relates to influencing your vendor. The more of those customers actively involved in communicating needs and desires to the vendor, the harder it is for the vendor not to pay attention. It’s really more of an issue of making sure there’s not apathy, and those customers are actually getting involved. That’s why I spend the time with MPE Forum.

Are you talking about apathy on HP’s part, as well as the 3000 customers’?

I find it incredibly foolish that HP ships boxes to customers without some kind of customer card you turn in. So you register your system and they know who their customers are. I find it entirely amazing that I had to bring this up to HP at a roundtable, and HP would reply, “Wow, that’s a good idea.” HP probably has a warehouse of calculators they’d love to give away as prizes in exchange for the cards. This comes under the heading of knowing your customer.