With network computers currently under attack by a number of alternatives, InfoWorld reporter Dan Briody went to get the story from Bob Gilbertson, CEO of Network Computing Devices, at ground zero of the NC movement.

InfoWorld: The NC landscape is very interesting and confusing right now and has changed a lot since NCD was founded. What is your impression of the current NC industry?

Gilbertson: The problem is, it's difficult to exactly define what the `NC' is yet because there are four or five different versions. There's the version which is the NC 1 specification, which we'll call sort of the fat NC because it has a browser and Java operating system on it. There's a JavaStation version, which is the one required for virtually all of the interface to be in Java. There's Bill Gates' new announcement called the Windows Terminal, which is as yet undefined, but [it] probably will be a GUI of a Windows-only [nature] that will allow you to do Windows. And then there's what we basically offer, which is a superset of all of those, which is a thin client that can talk to any server.

InfoWorld: There's been a lack of leadership in the NC marketplace. Does the NC industry need The Open Group to better define the reference profile, or is it better off with disparate types of machines and confusion in the marketplace?

Gilbertson: I think it's terrific for the user, both the user of the terminal and the customer who runs the terminal network, that there are alternatives. And in fact, the eventual winner will not come out of a standards group. [It will be] someone who delivers the most useful product when the customer needs it.

InfoWorld: In which markets do you see the NC being the strongest computing option?

Gilbertson: There are several markets. There's the dumb-terminal replacement market, which was the original intent. I believe that the network terminal/network computer concept works fine there, but it also works fine for PCs. It provides the function of a network computer as specified, and it also does the network terminal or Windows Terminal. So it's got a bigger market than anybody expected and it's a market that crosses a lot of different desktops.

InfoWorld: Are there bandwidth constraints with the NC? Does the number of users on the network affect the speed of a machine?

Gilbertson: If you properly design any network, it should work fine. The parameter we use is about 15 to 25 users per PC processor. On a 10MB LAN you'll never see any bandwidth constraints or slow down of performance unless all 10 users decide they want to do multimedia at once and [they] start downloading MPEG files. All you're passing is a very simple bit stream, which is painting the screen.

InfoWorld: There are some hidden costs in the network-computing model that are not talked about much. For example, don't the servers need to be very powerful and isn't some of the software expensive and sophisticated?

Gilbertson: There are some real costs that you have to put in place to build a suite of network computers properly, but they aren't that significant. Most people right now are buying desktops in the [price] range of $2,500 to $3,000: Pentium 166 MHz with MMX with 3GB to 4GB drives, big CD-ROM, probably 32MB of memory. For an NC, a rule of thumb is that you need about 8MB of memory per user on a server. So for 15 users you'd need about 128MB. You need four times as much memory on the server as you normally would at a dedicated desktop, but you're serving 15 people.

InfoWorld: What is NCD's thin-client strategy?

Gilbertson: We're going to offer a variety of products. It's unlikely we'll ever offer a product that has a disk drive. There's no reason to use a disk drive. We'll offer what we believe is the ultimate product, [a] universal network computer that has emulators to talk to every server out there. It has software that sits on a lot of these servers to help that server manage and identify its servicing-NCs. But it is unlikely that we'll get into the NetPC business. We don't think that model makes sense.

InfoWorld: How critical is Java to the NC model?

Gilbertson: I think Java is the thirtieth or fortieth language I've seen. It's very, very popular, everybody wants to learn it, it's very nice to do things that move from computer to computer. I've yet to see a good -- what I'd call mainstream, either desktop or BackOffice -- application that uses Java. Until I see that, I'm going to be a little concerned that it has general-purpose usage.

InfoWorld: Where does smart-card technology fit into NCD's plans and the NC structure?

Gilbertson: Smart cards are going to make the ubiquity of network computing happen by the year 2005. What we will do instead of carrying a notebook[is] we'll carry a smart card and you'll just stick your smart card in and it will do two things, basically three. It will identify what servers around the world you can access, it will have your password to verify that you are yourself, and it will have the billing codes to tell how to bill you. We think smart cards are really important.

InfoWorld: Are there any plans for mobile NCs?

Gilbertson: It depends on what you want to do with mobility. Right now I think it isn't viable. As technology improves and as ADSL rolls out to homes, we will get more wireless communication. Then I think remote network computing is probable.

Source: InfoWorld