sy to plug and play. Novell has to replicate both of those functions in the TCP/IP world.
BYTE:
Will IPX [Novell's proprietary LAN protocol] go away?
Schmidt:
[NetWare will] really be transport-independent. We have such a large installed base that we'll ship IPX in a legacy mode forever. The issue really is how do
you get all of your innovation, all of your new stuff on top of IP... and it'll be truly native IP, not encapsulated.
Our second main technology direction is our directory service, NDS [Novell Directory Services]. It's much more scalable than any of its competitors, and it needs to become even more scalable and to be cross-platform. We're introducing... a version of NDS called NDS on NT. We intend to adopt ADSI, Microsoft's internal interface to Active Directory. Part of our strategy is to interoperate with everybody else.
BYTE:
Microsoft has adopted a "surround NetWare" strategy. Are you going to take the reverse tack -- surround NT?
Schmidt:
Yes, that's exactly what we're going to do. Somebody sent me a motto: "We don't make NT, we fix it."... I actually think the future of Novell is not so much as a platform company but rather with the services that are enabled on top of NT and NetWare... If we can offer services that run on every platform and solve
some problems like network management, software distribution, or network integration, or auditing -- that category of problem -- then we become a very strategic vendor for those companies.
BYTE:
Won't people be interested in servers that are extendable, that they can modify, or that can be modified by other applications? Where does NetWare fit into that world?
Schmidt:
There's every reason to believe that the NetWare model can be the leader in the server-side market. The reason is that we have more control over performance and latency than the other guys do because we do not have a general-purpose kernel. We can present the very best Java engine because we can embed the Java engine in the kernel and the Unix and NT people cannot.
We're in the saddle point as a company. The NLM [NetWare loadable module] architecture is not very strong anymore. Java is going to take off. The question is: what year and how fast? We have a kernel that's part of
Moab, the next version of NetWare. It has controllable preemption and non-preemption. You can really control protocol latency. And it is transport-independent and has full symmetric processing and full virtual memory. It has very little assembler code and is written in C. It has a new file system, and a Java virtual machine inside it. Other guys have to do round-robin scheduling for Java. We have a real-time OS for protocols.
BYTE:
You're placing a lot of emphasis on Java as your development solution.
Schmidt:
We're betting the company on it.
photo_link (46 Kbytes)
Where's NetWare Headed?
