On the Future of Thin Client/Server Computing

<I>ENT</I> editors Thomas Sullivan and Al Gillen recently spoke with Ed Iacobucci, founder, CTO and vice president of strategy and technology at Citrix Systems Inc., about how thin client/server computing will work its way into the corporate networks of the future.

The thin client/server computing architecture is making a comeback of sorts. With the release of Microsoft Corp.’s Windows NT server 4.0, Terminal Server Edition, the half-dozen Windows-based terminals that support it, and the emergence of Java-based clients, the resurgence seems well on its way. Citrix Systems Inc.’s independent computing architecture (ICA), the protocol that enables thin clients of varying flavors to connect to servers, is an integral part of the progress. ENT editors Thomas Sullivan and Al Gillen recently spoke with Ed Iacobucci, founder, CTO and vice president of strategy and technology at Citrix Systems Inc., about how thin client/server computing will work its way into the corporate networks of the future.

How big is the thin client computing market?

That’s a question we’ve artfully dodged for many years because we didn’t really have a sense in terms of what the top end was. Today, if you look at the collection of companies that are working together in the thin client/server marketplace, it is roughly about $1 billion for 1998, including all the various pieces of the complete solution and the services and products that are delivered. We believe that by the end of 2001 it will be a $10 billion market and, in fact, every time we look it seems to get a little bigger.

Who is the typical thin client user company?

First of all, let me just correct one thing: We talk about thin client/server computing, and everybody forgets the server portion. The thin part of it has caught on, but in reality it’s about server-based computing supporting arbitrary devices, including thin devices. Most of our installations, in fact, are not only thin clients; there’s a broad range of devices and network topologies. About 50 percent of our clients are not thin clients at all, they’re the latest Windows environment combined with Macintosh, Unix systems, DOS, and thin devices.

Does sufficient industry backing exist to make thin clients pervasive?

There really is starting to be, and we’re starting to see different types of players emerge as leading participants. Of course the big guys are in both the thin client and PC markets. Through our relationship with IBM, they’re putting the independent computing architecture technology in all their NetStations, along with Sun and Java. These technologies enable a thin client/server architecture. But, also, if you look inside the thin client/server worlds, you’ll find companies like Wyse, Tektronix, and some of the Motorola divisions that are making thin devices.

Will there ever be a place for Java stations in the NT thin client market?

Java is a wonderful environment and it has enormous support within various camps in our industry. But as a device architecture, it’s got the same flaws that Windows has in that it is local execution, and with local execution you have to maintain an environment that can run the applications. As the applications grow, the devices obsolete themselves, which is the number one issue that IP customers are trying to circumvent by going to a fixed presentation services model.

I don’t think Java will be an alternative to thin client/server computing. But thin client/server computing augments what people are trying to do with Java in that you can run applications that are server-based and make them equally accessible from the Java device as you would from a Macintosh, Unix, DOS, or Windows system.

You mentioned earlier that you are working with IBM and Sun to integrate Metaframe technology into their different Java network computers.

Metaframe is a product, it’s the brand that we market. But behind that product is the independent computing architecture. It started out as a simple line protocol for devices kind of like a Windows system, but it evolved into an architecture for managing the servers, for providing end-to-end connectivity and manageability, and for scaling the clusters. It’s basically all the things traditionally in a mainframe data center being pulled together in a consistent architecture where the pieces can be developed and fit in.

Is it safe to say that the Java Virtual Machine those machines use will become outdated in a short period of time, but that it really doesn’t matter if you are running Windows applications using independent computing architecture?

No, it doesn’t matter at all, but it takes a little different thinking. We’ve all grown up in the PC industry -- and I was a participant in that industry all through my years at IBM and Microsoft -- but what we’ve always said as a PC industry is that you can scale up the hardware and the software is going to scale and so forth. The problem in IT organizations is that it’s easier said than done. If you want to roll out a sophisticated new application it’s becoming increasingly difficult, and in some cases impossible, to upgrade the entire infrastructure in order to roll out a new version of the applications.

The way we like to talk about it, it’s the hardware and software upgrade cycles that are inexorably linked. Now, one of the things that we do is provide a robust set of functions right around the server with a protocol that delivers them effectively. With this you can build function stations that don’t become obsolete as the applications evolve and you can more easily upgrade infrastructure.

How do you think thin clients will be defined in the future?

We have a very specific definition that talks about 100 percent server execution, and that’s a very important element of a thin client. Anything that runs application code as a client is by definition not a thin client because it isn’t a matter of thinness or thickness, the name is bad. It’s about the same piece of hardware or the same piece of software running on a workstation accessing the same set of servers and it doesn’t matter what kind of applications you use, you don’t have to upgrade the workstation to keep up with the applications.

What impact will Windows 2000 have on thin client/server computing?

We’re hoping it’s the next evolutionary step. We’re pretty actively engaged in developing server-based computing code. We signed an agreement with Microsoft where we licensed to them some of our server technology that makes NT into fundamentally a multiuser system, which is like the entry point. You have to have that level of capability before you can build a server-based computing solution. So we licensed that back to Microsoft, and we have an ongoing joint development agreement where our engineers at Redmond are actually doing all the integration for the server-based multiuser capabilities of NT. We’re pretty excited about the next level of functionality that we’re going to be able to deliver to that platform, and we’ve got companion products that are going to be delivered in conjunction with that. So yes, I think it’s another step in terms of function, if you will.

How do you see the thin client market changing in the next five years?

Five years from now we’ll be using thin clients, but we won’t even know that we’re using thin clients. For some examples, wireless handsets that you can call and turn on three way calling and turn off three way calling. Only you’ll be able to call your mobile provider or maybe even go through the screen and turn on and turn off different types of applications that you want to run. Now this isn’t Excel, Power Point, Lotus Notes or anything like that. But it will be things like e-mail, the Zagat’s restaurant guide, maybe a yellow pages, maybe an on-line travel agency, stock quotes. A lot of this stuff is starting to happen today, but once there’s a well-developed, well-deployed architecture that allows IT to be creative, in much the same way that Microsoft allows them to be creative on Windows, we are going to see a plethora of different types of applications that you can deploy, not just on the wireless products, but also on set-top boxes, ultimately on television sets. You don’t even need a set-top box because a fixed architecture like ICA can be embedded in a single chip and never have to be upgraded. It can be part of a digital television set. We’ll be seeing all sorts of things lined up as new services, with a services showcase, but underneath it all will be a thin client/server infrastructure supporting it.

Alicia Costanza contributed to this report.