Harry Sterling Q&A:Managing the 3000's Growing Pains
Click here for HP Sponsor Message

NewsWire Q&A

Harry Sterling
Managing the 3000's Growing Pains

In part two of our interview with CSY General Manager Harry Sterling, we hear about encouraging application providers to move to the 3000, retaining Oracle as an MPE solution, defining New customers for the 3000, the promise of Shareplex, and improving IMAGE/SQL and native 3000 Internet services. Plus, Sterling's biggest surprise in his first year of managing the 3000 division.

It looks like the strategy for the HP 3000 Grow customers, for the moment, is missing a couple of elements: the IA-64 promise for the 3000, and a way to encourage development of applications for the 3000. Both of these are solved for a Supplement customer. Of course for a migration customer they might be solved too, but nobody at HP wants to talk about migration customers for very long.

Well, it's not really migration, and that's where some people get confused. It's really where the customer determines that their current application is not meeting their business needs. And manufacturing is an area we're seeing a lot of this, and they need to replace the application.

So they're migrating their application.

No, they're not. Migrating means taking what you have and moving it to a different platform. Replacement says, "I have this accounts payable package that is batch oriented, doesn't have the front-end interfaces with PC tools, and my people could be a lot more productive. Or it doesn't handle multiple currencies. I've become a global organization. I've got to replace this application."

Migration says, "We've got this application; we want to move it to UNIX because we think we need to be on UNIX, and so we want to be able to figure out how to move the database, how to recompile all our programs on UNIX and salvage everything that we have." That's not the model. And one of our strategic areas is not to move people for no good reason.

They wind up moving when they can't find an application.

They change platforms because they're replacing an application. When we've talked to a lot of our customers they say, "I don't want to move what I have. It's obsolete, I want to replace it." And so for us to make lots of investments in, for example, putting Image on UNIX or putting Vplus on UNIX -- when we talked to these customers they said, "No, that's not something we want to do. We're happy with the 3000, if the application is meeting our business need. If we need to go to a UNIX box or to an NT box, it's driven by the need of the application."

There are a few theoretical situations where very high-level executives say, "We're going to move everything to UNIX." Northern Telecom was one of these situations. I talked to one of their VPs and said, "I'd really like to understand, what's your business case for doing this." And he thought about it and he said, "You know, I don't really have one." A couple of months later they decided to modify their strategy. Their manufacturing people rebelled and said, "There's no reason for us to reinvest in the hardware and to do all of this re-engineering of existing applications that are meeting our needs. There's no business driver to do that." And so now they're basically a Supplement customer, and they're moving to UNIX when they need to replace an application. Customers have come to that realization that just the underlying platform is not a business driver.

I think there's also a misunderstanding about what we mean by growth. We want to ensure that customers who have made huge software investments on the 3000 who need to continue to grow, can expand the performance capabilities because it's driven by the business growth. If they need to add more users or a second site, we want to make sure they can do that on the 3000 platform -- that they don't have to change platforms if their application is meeting their need. That's our definition of growth.

It means that if they want to implement a new application around the one they have -- they want to bring in-house a new distribution system -- it might be nice if that distribution package was run on the 3000, what we're saying is, it doesn't have to be. If you can't find one -- and there probably aren't a lot of really good ones --put it on whatever box the vendor recommends. Hopefully, it's an HP UNIX or an NT server, and we will make sure it works with your other 3000 applications.

We've been down this road many times about the applications on the 3000. I would love it if we had more applications on the 3000, and as I've told you before, we've spent literally millions of dollars paying vendors to court them. SAP is a good example. They actually had R3 running on the 3000 along with many other platforms. Once they started implementation and started revving their software releases, they recognized they could not support all of the platforms that they had originally intended to support. They had to scale back the numbers of platforms they were willing to support, and the 3000 was one of the ones they cut -- even though we had spent millions of dollars in Boeblingen, Germany, we had engineers helping them do the port.

So paying them to come to the platform doesn't work, because it comes down to again, a business model. From their perspective they ask "Where can I gain the most leverage with the least cost to my organization?" It's really funny, with a new start-up solution company, they want to put it on everything to start because they want to have the maximum exposure. But when they get down to the harsh reality of going out with their revisions and the support, it suddenly hits them right in the face that "We can't do this. We can't re-certify on 15 different hardware platforms. We can't maintain 15 versions of our product and do 15 sets of patches." They scale back to two or three, depending on the size of their organizations. And the 3000 is not one of those.

You've said you know it's pointless to throw money at software companies to get them to do something that they can't support in a business model. What else can your division do from an HP leadership level to encourage applications providers to offer 3000 solutions?

To get them on the 3000, literally nothing. I mean, we've pursued this. We made huge investments in Posix -- that was our original intention. I'm not saying that it was the wrong thing to do, because it's paid off for us in many other ways.

The original intention was to get new applications, and we've tried to make it as easy as porting to another UNIX platform, but it didn't fly. Because of this issue from the vendor's perspective, they can't support all these platforms.

And the same thing happened on the UNIX side, by the way. They pulled off with NCR; they pulled off of you know, they picked the one or two key UNIX vendors, and now picking NT as one of their platforms, and that's it. They're not willing to support any more than that.

So in looking at it from that perspective, we concluded there really was not anything we could do -- because we even tried paying them to come. WNow the one vendor who has been very loyal and has followed through is Oracle. They've done a great job. They're continuing to do a great job. Still committed to the 3000, and so that's been a very strong relationship, and they feel they've gotten a significant return on their investment from the Oracle applications that customers have bought on a 3000.

What can you say to Oracle to make them do Oracle8 for MPE/iX?

Again, it's going to come down to a business case.

But now that you you're learned that, can you help make a business case for them to continue Oracle8 on the 3000?

It's going to be our customers that decide in terms of how well they continue to adopt it, how much they continue to buy. If there's no return for Oracle to do this -- I mean if the funnel dries up -- then it's going to go away.

How proactive do you want to be?

I think all along we've talked about the strength of Oracle. It's in every one of our pitches how strategic Oracle is, how well we work together, so I think it's very important.

Can you do some kind of a summit meeting that you can arrange between some of your biggest accounts and someone high up at Oracle -- to get these customers to explain, "If you do 8, we're staying with it, and we're recommending it for our other divisions."

If you really take a look at our supplement strategy in customers who want to implement -- and this is another strong case in support of Oracle gateways -- as they implement new applications on UNIX with Oracle, and the fact that it works very well with IMAGE is giving them some investment return on the Oracle gateway side for the 3000. So even as customers move into the supplement strategy, and in some cases, if they want to deploy an application on the 3000, Oracle is a good alternative for them to be in and then move more towards a UNIX environment if they want to do that longer term or NT, and so Oracle--the Oracle environment gives them a lot of flexibility to do that.

I want to go back to the first part of your question, which was about growth and make sure you understand the difference. It's not the growth of the 3000 business, it's growth opportunity for customers' current software investment. It's their business growth. We want to be able to support their business growth -- not growing the 3000 business with new applications on the 3000.

So that particular phrase doesn't address the size of the 3000 investment, it addresses the size of their business operations.

For a customer who is 3000-centric and is a growing business, we want to make sure we still meet their needs five years from now. It doesn't mean that they can get new 3000 applications on the 3000. For those customers who have their own applications, we want to make sure that they can continue using the 3000 as long as the application's still meeting their need.

Are the HP 3000 Grow customers going to have fewer options because they're not going to be investing in HP-UX systems?

Yes, if they don't get off-center of looking at it as a 3000 world, they're going to be limiting their solution options, absolutely. As a matter of fact, that's why we've really put a lot of effort into the supplement strategy. The last time we looked, about 45 percent of our active customers are actually in the supplement segment, so the majority of our active customers are already in the supplement segment. They have concluded that deploying new applications on other servers and co-existing with their core 3000 applications is working for them. That's where the majority of our investment is going -- for things around middleware and networking.

The growth segment is focused on performance and supporting new chips. Effectively that's all we're doing with that segment. We're going to make sure they're going to have a bigger box. In a few cases, because of their business growth, we know that even with the bigger box, we're not going to meet their need, which is why we're looking at the horizontal growth option.

Several channel partners have asked me, "How's come there wasn't a New customer segment to go along with Grow and Supplement?"

We have new customers in all of these segments.

But why not identify it as a type of customer?

Because that's the wrong way to look at the business problem.

So everybody that comes in is going to be in one of these categories. Well, not many of them that come in will be in the replace category.

Actually, a few have. They're replacing other applications. We've got some who have replaced AS/400s with 3000s. I mean in that situation, they looked at the applications and decided that the 3000 version was better than the AS400 version they were running, and they replaced it. So that's what we mean by Replace. That [New segement question] is kind of like the old technology way of looking at the problem rather than looking at it from a customer environment.

Yes, it's the channel partners that raised the question.

We're trying to look at it from a customer's environment point-of-view. Even in some of our smallest users, they've got Netservers in there, supporting Notes or e-mail, and they want to start linking those with the 3000 solutions. I went to one small company where they had all of this information about their manufacturing on the 3000. And there was a quality department that used to take reports off of the 3000 and then re-enter the data on the Netserver PC. All they needed was a link between the Netserver and the 3000 to be able to do the extract of the data.

They are in a replace segment, but they've separated those into two separate environments, and they're looking at the 3000 piece as 3000 work, they're looking at the NT piece as the workgroup's responsibility. What we're trying to say is, "Hey, let's bring all of this stuff together and make your organization more productive."

There's some customers who really -- if you look at their environment -- would would be identified in the supplement strategy. But they self-identify as two separate groups, an NT server group and another is a 3000.

That has some implications in the complexity of management. It doesn't really make any difference how hard you work on it, a distributed environment is never going to be as simple to manage as a single box.

The reality is that there are not a lot of other options, because we're still very competitive on the high end. We are with UNIX, we are well into the mainframe arena with IBM. So if their business is growing that quickly, then you know, there's not a whole of choices they have but to look at other architectures to deploy.

So in addition to redundancy and availability advantages of Shareplex, now you'd like them to look at it from a performance standpoint?

We're doing work with Amisys and Smith-Gardner. They're they're changing their applications, we're doing work in performance modeling, we're making changes to FDDI drivers, for example, supporting more of them so you get more parallelism. We're really focusing in that area, and that's driven by the growth segment -- customers who had existing applications that need to grow and their business is growing more rapidly than the traditional growth in chips, regardless of whose platform you look at. They need a different computing model to meet that business growth need.

Do you think that CSY has a role to play in encouraging small software providers to create applications for the 3000? Is there an opportunity out there for software companies to hit niche markets? Or do they all have to go and try and pony up to the biggest market that they can possibly touch with the most resources?

If they were trying to build an application -- some kind of an application, I would think that particularly with that small one- or two-person company that one of the things that would be really critical for them would be to be able to sell a total solution at a very low cost. I would think that a better solution would be a Netserver or a PC solution.

I mean, you've got penetration: a small operation like that is not going to have the infrastructure to support a Fortune 500 company. So you're typically wanting to play with the smaller companies, and probably locally. You're not going to find 3000s, or if you do, they're going to be not willing to upgrade or buy a new box for that application. They'd be more interested and you'd be more competitive with your solution if it were NT-based. So, I guess if I were doing this, if I left HP and decided I wanted to start a little software company, for example, I would not choose a 9000 or a 3000. I would choose a low-cost box.

So that the hardware wasn't an impediment to your sale.

Right. The 3000 and 9000 servers are really targeted for the enterprise. Even today with some of our long-established third parties on the 3000, some of our Fortune 500 companies are very nervous about relying on a small company for a major piece of their solution.

That definition of small just keeps getting bigger and bigger.

Yes, and that's where I think the way Bradmark supports [its spinoff Starvision] is a good model for them to get little contract people. They basically are the front end for offering the solutions. For somebody who wanted to do that, that's the way to do it.

I still have to say that for an application, I don't think you'd choose the 3000. I think if you were trying to target a tool -- for example, the stuff that Paul Wang does in SolutionSoft -- he's got a great little niche market there and he understands the 3000 very well. But if he decided he was going to come out with a payroll package, I don't think he could play it with the 3000.

So do you think the 3000 is a strong enough database server then to maintain its place in the enterprise without new applications? One of its strengths is being able to serve data and move it back and forth from one box to another.

Its strength is in leveraging existing environments and applications. In the key solutions we have on the high-end like Smith-Gardner and Amisys, where we get a lot of new business, that's where it's really strong. But for the small- to medium-sized business looking at new applications, it's basically an expensive solution when you compare what's going to be happening. If you look out four or five years, it's hard to compete with a Netserver solution for a small- to medium business. And with HP's new architecture in the future, we'll be able to offer the hardware platforms from the desktop to the enterprise based on NT.

Why hasn't Domain Name Service for the HP 3000 hasn't been a higher priority yet in offering the 3000 as an Internet solution? If the 3000 is going to stand out on the Internet as a full first-class citizen, why does it need to rely on another box for DNS?

Looking at the Internet and the Web servers within the supplement solution -- which is where I put it, because the Internet by definition puts you in that segment -- the functionality already exists on another platform, and I'm not going to move it to the 3000. I'm going to put on the 3000 only what needs to be on the 3000 to allow you to interconnect into that Supplement segment.

My investment is going to be focused on making sure the 3000 works really well, high performance, high reliability in that server -- or in that supplement space. If you want a firewall or a Domain Name Server, you look at putting in a Netserver or a UNIX system to provide that functionality.

A system that's less reliable than an HP 3000.

But the whole Internet is that. The 3000 focus is going to be on the database information that you want to plug into the Internet, okay? And that's where I'm going to use my critical resource. I'm not going to waste time duplicating functionality if it can be provided in some other way.

Now if I were looking at this from the 3000-centric point of view, with the definition of growth that says you will be able to do anything you want on the 3000, then yeah, I would use that other model. But the Web is in the Supplement segment -- that's the way we're looking at it. So assuming your 3000 is going to be interconnected into a Web of UNIX and NT systems, if that service or functionality exists already on one of those servers, I'm not going to duplicate it on the 3000.

Why does IMAGE/SQL have to maintain its reliance on Allbase/SQL? Hasn't the market changed enough that IMAGE deserves its own native SQL interface?

I'm not in the database business. I'm not in the business of trying to make IMAGE/SQL a strategic database player in the future.

What if Oracle decides not to bring Oracle8 to MPE/iX? Does that change anything for you?

I will continue to provide functionality in IMAGE/SQL driven by the needs of the customers for the installed base applications.

If they want it bad enough, they'll get it. I mean, that's kind of been the philosophy for the last years.

But that doesn't say I'm going to rewrite IMAGE/SQL in order to eliminate the dependence on Allbase. That would be a stupid waste of resources when I could use those resources to work on something that's more critical, that would give me a bigger return on the investment. If I were to do it, I would not get any more sales for the 3000, or any more upgrades, or add any new functionality that would cause customers to install new HP systems.

Do you think it would have an impact on satisfaction for the existing customer base? Would people be happier with the ease of use in administering their databases, or aren't there problems out there with this interface?

I understand that. If it required me rewriting IMAGE and it took me 10 engineer-years to do, is that the way you want me to spend my resources? Or would you rather have me work on greater than 4 Gb files. I mean, it's a finite set of resources.

You've been doing your job for more than a year now. What's been your biggest surprise?

I guess the biggest surprise is that "the buck stops here." There isn't anybody to go to for advice. It's me. I mean, I have to make the final decisions that relate to the 3000 business. I don't go to Glenn [Osaka, his direct report] and say, "What should I do?" You know, when I was the [CSY] R&D manager, I could always bounce the ideas off of the GM or talk to the marketing manager. I still talk to [marketing manager] Cathy Fitzgerald, talk to [R&D manager] Winston Prather. But the point is that I am the ultimate -- I have to make the final decision about the 3000 business. So I guess that that reality kind of has hit home. You know, it's kind of like, "Who do I need to ask about this?" And the answer is, "Me."

It's funny you should mention that because Glenn told a very similar story in August. He had the same experience when he got into his position. It was like, "Oh I get it. I get to make the decisions."

Right. I guess in every other position I've been in there's always been, you know, at least one person above me who was ultimately responsible for the final decision related to that business that could override it. That's not to say that if I did absolutely something stupid Glen wouldn't step in. For example, if I don't meet my profit objectives for a quarter, then suddenly I get somebody's attention -- which I don't want to get.

You know, being invisible to the upper levels is goodness, because when they get involved it's because something's not working. When I get [CEO] Lew Platt letters, or escalation issues from [Computer Systems Organization GM] Dick Watts about an unhappy customer, he sends it to me, and that's the end of it. He knows that I will deal with it. Sometimes he'll say, "Well, let me know what the resolution was," you know, and it may take a month to resolve, and I'll send him back, saying, "You know, that issue has been closed or whatever." But he doesn't get involved in the details of it.

If I have to make some decisions about financial compensation or if I get involved in legal situations, I have to make the decisions. Me. I don't go up and say to Glen, "What do you think I ought to do? Here is my recommendation. What do you think?" I mean, that step doesn't occur. It's like, I have that conversation with myself, going back and forth to work.

From the look on your face, I would say this comes as a good surprise.

It's sometimes a lonely place to be.

Now that you're getting to make those decisions, how do you feel your management choices have improved the 3000's value to the customers during the last year?

I can measure that based on the pulse of the customers and what they're saying -- how they're feeling about what we're doing and the direction we're going. From what I'm hearing, it still seems to be very positive, and so, I mean, I can't look at any one decision that I make and say, "Was that the right thing to do or not?" I have to kind of look back over the last year and collectively say, "Did we meet our profit objectives? Absolutely. Did we meet our expense objectives? Absolutely. Did we retain the key people in the organization? Are they highly motivated? Did we get high scores on our employment surveys--our employee surveys?" And the answer to those are all yes.

Are the customers still happy, and are we getting a lot of complaints? I mean there are all different kinds of measures that you kind of can think about over the past 12 months, and I have to conclude that -- for the most part -- we're doing things the right way. From an HP perspective, we're successful. From a customer's perspective, I think that for the most part, they're pretty satisfied with what we're doing. Those are the two key measures for me.

Copyright 1997, The 3000 NewsWire. All rights reserved.