IDG: Will Novell be toast if NDS doesn't gain market acceptance?
Schmidt: Why put it in the negative? NDS is working. We have 40 million users of the direc-tory. There are approximately zero Microsoft Active Directory users. We have majority market share of NetWare against NT. Our unit sales are growing, as are NT Server's. NT is good in some things. We are good in others. It's called competition. But customers make choices based on what they can purchase today. If you wait for products to arrive, you may be waiting a long time.
How big an opportunity does NDS represent?
It's clearly as big as the SQL database opportu-nity of 15 years ago, which is now worth $US80 billion. The long-term principle is we build our business model around applications that use the directory. What kind of applications will be directory-enabled? What are the set of problems that you have on your network? You can't find things. You can't manage stuff. You can't do software distribution. You have no idea how many software licences you have. You can't find people. You can't do instant messaging. You can't do certificate revocations. All these functions change in the presence of the directory.
Are you frustrated by the fact that there is so much head-to-head comparison between NDS and Active Directory, which hasn't shipped yet?
Sure. It seems obvious to me. Let's compare shipping products to shipping products. When Active Directory ships - which I am sure it will - we will have a proper comparison.
You're fighting an amazing perception battle, the perception being you're losing ground to Microsoft. How do you combat that?
It takes a long time to change the conventional wisdom about a company that has been beaten up as badly as Novell has. I mean, how many quarters of rising revenue, rising profits, successful products would it take to get a positive story written about us? It takes a long time.
When I hear talk of users swapping NetWare for NT, it sounds to me more like a sales problem than anything else.
When we have sales coverage, we keep customers and they make big investments in us. In the historic channel model the only choice was Novell, so we didn't have to go in there and sell really hard. Now we do.
Is there urgency in terms of how quickly NDS ubiquity has to happen?
When I showed up I wanted everything done right away. But I quickly figured out that there is a rate at which organisations can do stuff. And until we shipped NetWare 5, no one would believe our plans for the Internet because they weren't tangible.
Now that we've shipped NetWare 5, we can go to companies that wouldn't look at us in the past because we didn't understand the need for interoperability that the Internet has forced. Now we can say that not only do we understand it, but let us show you how we work with your networks. It's now extremely easy to demonstrate interoperability with all of these other players because the native protocol is IP and all the services are layered.
We can now walk into the boardroom, do an effective demo, and say: "Let's do this deal now."
Why have you turned your marketing efforts upwards from your traditional network manager base to the CIO level?
That is a new thing for us. While we had this great channel, we didn't call on CIOs. And in the last few years, the buying decisions have been ele-vated from the systems guy to the CIO. The people we need to get to have either never used a Novell product or don't know why a directory service is useful.
Ultimately, it is not going to be profitability or revenue growth that will change the belief in our company among people at that level. Things will change when they see us driving a space in the market that no one else is driving. In the minds of the average user there are big gorillas - such as Intel, Microsoft, whomever - that occupy market spaces.
Finding a space and trying to innovate in it will be how we get noticed by CIOs.
We are going to do that by delivering directory-based applications that run on the Internet and change the way people think about identity. Once that is done, then these questions will get a lot easier for me. That will take the question about our credibility right off the table.
You once said ZEIDGorks was a good idea you found hiding in the labs. Have you found any other good ideas that can help move NDS along?
Every month I still find these little gems. For example, we have technology in our lab in India that is essentially a client-side directory. What it does is talk to the big directory and it has information about you and all of your preferences.
We are developing a set of caching appliances whose caching performance is a hundred times better than what anybody else has.
And I haven't even talked about the next version of the directory. Let's say that the current directory is more scalable than anyone else's and the next one is even more scalable than that. The way that is achieved is by changing the underlying data store.
We cull the company for these technologies on a monthly basis, and we just put them together.
Lucent recently agreed to adopt NDS. Is that deal important in itself, or more so as leverage to get other hardware vendors such as Cisco to sign on?
Lucent is clearly one of the two or three companies that will hold ground in the datacom market. Everything is going to consolidate, and Lucent will be one of the consolidators. Network infrastructure is a game of scale. Lucent has enormous scale.
The reason this deal is important is that it gets this notion of identity and manageability into the high end of Lucent's product line with their Cajun switch. That is where all of the new and interesting services get built first. And then the NDS integration will be propagated throughout the rest of their products.
Obviously, we would like to do similar deals with other vendors.
What about systems management? If you integrate the directory with other systems, don't you run into problems over which is going to be the master?
Let's think about the right architecture. At some level in networks everything is a manageable object.
Your e-mail is a manageable object, your router, your PBX, your desktop, your applications.
They are all manageable objects. Why is there not a single object data store that any set of management tools can tap into?
How many years is it going to take for an enterprise network to get to a single directory service?
It takes a while. The big customers we have are rapidly putting every PC on NDS - 20,000, 30,000 and 40,000 PCs. This is because once they have them all on NDS, they can count stuff. That is what they are really trying to do. The move to a single directory is basically driven by the need for asset management. It is not yet driven by manageability reasons; it is driven by cost control/ inventory kinds of reasons.
Every customer I talk to tells me their network costs are out of control because stuff doesn't work together: "I need something to manage all of the weird variants of clients"; "I need server-to-server management"; "I need to view everything as a single, coherent network and decide what goes where".
That is how they think, and that's what we're doing.