InformationWeek: The Business Value of Technology

B G: Any surprise about the 80% who said they plan to or already had some distributed systems? Jerry's shaking his head no.

JP: No, I'm not surprised at that. We've all said here that everybody goes through on the server strategies and what everybody learned out of that. People got burned in client-server. No matter what infrastructure you're building, at least put it in one place so you can keep a handle on it for all of those reasons.

CF: You know, there's a very important group of partners that we don't have here that will help answer that question. And that is the networking partners--be that the carriers themselves or the Ciscos and Bays, etc. With the advent of the Web, with higher-speed technologies and some promise of unembraced technologies like ATM, you have the opportunity there to say: "Well now, maybe the computer doesn't have to be out and distributed. Because I don't think it's that people wanted to have a bunch of distributed systems." Users these days don't want the computers out the re; they want the power out there. [When they've got that] they're going to do that and they're going to have all kinds of new applications. It's going to increase the network and the bandwidth requirements. And that part of our industry is going to have to work with us to keep up.

GR: It seems like several years ago that IT departments didn't necessarily want to lose control of the architecture of infrastructure, if you will, but the business units were being empowered to do things to make more money and so they took control of the asset. I think the question of whether it will come back to centralization again is still in the business unit's hands--that if the head of the organization hasn't been convinced that it's the right answer, I don't think it's going to happen.

DR: I guess there is one missing point that has really pushed the centralization , and that is the year 2000.

CP: The group alluded earlier to the people component, too--is that a factor for the users here--just having the righ t people resources? Is there a link there between the technology and the right set of people?

EB: I find that in our environment, it's hard to get qualified personnel where you are locally so we've actually advertised on the Web to get our personnel. And we've brought them in from Michigan just to get the quality because it's just not there. Distributed [computing] would make it harder to manage someone you don't meet, you don't touch, you don't feel and taste, and it's hard to get that person involved in the team to get the applications written, to get your systems out into the marketplace. The management of the systems themselves for us has to be maintained in one location because we are the brain trust of the organization knowing the applications and knowing how they have to be distributed throughout. But my mainframe, I'm not going to spend oodles of money. I've invested millions of dollars writing these systems and I'm not going to spend any more to rewrite them on Unix. I'll find a way. The pres entation is what [the users] are all intrigued with--and if you have the clicks and that point and shoot, then that's enough.

CC: I think stiff competition and a need for companies to focus on their customer is going to force them to have to standardize internally in their systems. I don't think companies these days can afford to be running 20 different systems that have 20 different customer codes for the same customer. And I think the move to standardize the data will force standardization of applications and that, in turn, will drive some future centralization of hardware platforms.

KM: When I look at our customers and ourselves in particular, what we've done, we had in excess of 30 data centers at Unisys at one time. We've got that down to one as of about two years ago. But one of the keys of being successful is not removing the choice and the freedom for people to use the software or the applications the y need. People still want to have the flexibility. So when you can bind down to a single one, you have to make sure that you're still able to offer your users, your end user, whatever addresses their particular needs. If you can't offer a system or a host of systems that can give them, whether it's an NT-based, legacy-based or a Unix-based application, I think you'll find that most people slowly rebuild their little islands.

CP: Is it fair to characterize that we're headed towards a situation where we have a centralized architecture, even a centralized culture, but a distributed look and feel? It sounded like that's what you were saying.

KM: Absolutely. With the browser and front ends having access to the application. Everybody wants to point and click. What it runs on really becomes inconsequential. If you're the end user, all you care about is: I've accessed what I wanted in a timely fashion. It doesn't go down. It's backed up for me because, let's face it, the worst everybody does is that nobody backs up or very few people back up.

CM: It seems like in the '80s it was all centralized in green-screen sort of environments essentially. Then, distributed computing came along with the advent of computer power going out to the end user and everybody wanted freedom from the data center. And so you got this rush to freedom and autonomy. I don't think the data-center management reacted very well to it. They just let that happen. Now, we've got the technologies that provide distributed computing but recentralizing data. So I think the pendulum has stabilized a bit. I don't think you'll ever take complete power out of the individual users ... I see more of a recentralization of the data rather than the computing.

CP: Let's move to the Web as all good discussions must. Do you consider your mainframe server as a Web server? Some 41% of the people said yes and 59% said no. [Do you have plans] for Java-based development for mainframe or large enterprise-class systems? [Of our respondents] 45% said yes while 55% had no plans. I'd love to hear from the users here what kind of Java development is going on in your organizations.

EB: I'm part of that 54.4% [with no Java development plans for enterprise systems].

CC: We have some Java development going on. Obviously, Pepsico has Web sites available. Pepsi-Cola has a particularly entertaining site targeted at teenagers and [people in their] early 20s where there's a lot of Java development done. So a lot of the development that's been done has been for the consumer who is out on the Web looking around. In terms of electronic commerce types of business applications, there's certainly plans to proceed forward. Do we have anything that's deliverable at this point? No. But there's some development efforts going on. They tend not to be on mainframe or enterprise-class servers. They tend to be NT-based servers at this point.

MP: I need to broaden the definition of mainframe again and say enterprise server because we don't have mainframe. In the short term, they're not doing much--but Java is very important to us because we're really using fourth-generation products and third-party products that are developed that generate Java or were developed in Java to deploy our mission-critical assets. Java is a very critical thing to us, even though we're not doing a lot of Java development in-house. Our plans to do development of Java kind of got set back by one of our key software suppliers, Oracle, who didn't release the capability to write store procedures to Java with the Oracle, I guess 8.1 or whatever.

GR: We're not doing any Java right now. We're looking at it. We're a very Microsoft-centric organization and we wouldn't do it on a mainframe, we would do it on a Unix or NT server.

JP: If you're really running your heartbeat applications and they're doing your business transaction processing on a day-to-day basis, for Ja va to be of use there, it has to somehow integrate into that transaction environment whether it be CICS or Tuxedo or whatever.

GW: It's interesting. We find customer reaction very different--at least the customers that I've been talking with. There's a lot of interest in using Java in the System 390 world. Just about month or so ago, we announced the Component Broker series, which basically is objects with transactional integrity and mainframe-class kind of characteristics that can be evoked out of a Java environment. We have so many customers that are interested in that, we can't handle the number. They're looking to deploy their assets in a way that they could connect to the Web world. And without the transactional integrity, etc., that we could offer, they couldn't really do that very well.

CM: Java ... [is] a guaranteed success. Sun invented it and IBM jumps on the bandwagon and supports it and Microsoft will eventually claim credit. So it's a guaranteed success. I know of no independent softwar e vendor that's not doing Java development in some way, shape or form. And I would like to meet one. Are they right in the core of their product--is Oracle right in the core of their database engine in it? No, not yet. But everybody is doing user interfaces and defining management environments and Web hierarchies and all those sorts of things. So I'd be surprised, even the people that say no, that they have no plans. I'd be surprised if they're not doing something on it, even through a third-party software package. So I'd see it as taken over from an interface point of view on how the primary vehicle that users will use to access the data and the applications. I think it'll be awhile before the key production codes are rewritten in Java.

JS: That was my question. Are the people that say they're not doing Java work on the mainframe--are they using Java for even Web-based interfaces to access this enterprise data? I see it quite a bit.

CP: Do you plan to have network computing access to the mainframe ? We're really trying to blend NC's here with the mainframe: 68% said yes, that they plan to and 26% said they are not planning to.

MP: The important thing here is thin-client computing, Web computing or whatever, not the hardware device. I think too often they get mixed up in discussions. Because it doesn't make any different whether it's a Net PC, a PC with a browser or an NC on a desktop. We're definitely deploying NCs for internal use. That's practically all we're deploying today in terms of new workstations. It's probably nine out of 10 new purchases. And a lot of it is Unix applications as opposed to PCs, although we're finding a lot of people who have a PC end up giving up their PC voluntarily because you really can provide enough functionality through the NC environment to give you some access to the NT world.

BG: By the end of this year and by the end of '98, how many NCs will be operating?

MP: I haven't really got a good count of what there is. Probably a couple hundred there right now . By the end of this year, there will probably be a couple hundred more. The big thing will be when we put it out in stores. There's 250 stores.

CF: One trend we see towards the thin-client processing paradigm is security and control. The second thing that we see is cost. We launched our total cost of ownership initiative thinking that we're going to lower the total cost of ownership. That's not what excited the customers: What excited them is the ability to have centralized management control the application and security. Then our customers said: "Hey, look how we can recapture all this desktop investment and extend the life of these devices."

Return to page 1 or
continue on to page 3

Back to News in Review

Send Us Your Feedback

Top of the Page