In the tiny, cramped dressing room named for Jerry Garcia in the bowels of San Francisco's Warfield Theatre, Microsoft president and CEO Steve Ballmer looks like the proverbial caged lion. At Microsoft's Enterprise Server 2000 launch event recently, Ballmer positively roared that the software giant, which enjoys a desktop monopoly so vast a federal judge branded it illegal, is ready to ride Windows into the heart of the enterprise. With the release of Windows 2000 in February, and the release of Windows 2000 Datacenter, the company appears poised to make good on the promises to prove the operating system high-end ready.
Leaning back in his folding chair, waving his arms and occasionally speaking in a voice that no doubt rang throughout the theater's basement, Ballmer sat down with InfoWorld Editor at Large Bob Trott and Senior Writer Tom Sullivan to discuss Datacenter and Microsoft's .NET "software as a service" initiative. Ballmer also made no bones about who is in Microsoft's sights: archrivals Oracle and Sun.
InfoWorld: This launch marks a big day for Microsoft's .NET strategy, doesn't it?
Today is several different things. No. 1, it's a big product launch, and product launches are always fun. No. 2, we're making a milestone in terms of our ability to serve the enterprise. If you take a look at the way our partnerships have evolved -- the 7,000-plus people who work in enterprise consulting and support, our ability to look a customer in the eye and say we'll put skin in the game on a big enterprise design win, and the customer references that we have, whether it's Home Shopping Network or Honeywell or electon.com or Dell or Pitney Bowes -- we have exciting customers talking about the way they're using these technologies. No. 3 is these are really the first products to be released with some .NET technology in them. That's really pretty exciting for us. It marks a milestone. No. 4, it's fun to ratchet the competition up with Sun and Oracle to another level.
How does Datacenter fit into the .NET strategy, other than being the highest-end server that you have?
I would say the first version has a little bit of .NET technology, and then we add more. Take a look at what we did in just Windows 2000, the way in which XML and managing of XML was super nice. Datacenter inherits all of that. We do need a scale-up as well as a scale-out approach. Datacenter does help with that. There'll be another round of Windows releases that supports the so-called .NET frameworks into Windows. The first technologies in that are really .NET technologies. You see it in SQL, in Commerce Server, certainly in BizTalk, and the Mobile Internet Information Server.
One of Sun's criticisms is that Microsoft is just catching up to where they have been for a while in the high end. Is there any validity to that view?
When are they going to catch up to us? That's what I want to know. When are they going to have a scale-out strategy? I'm looking for their scale-out strategy; I haven't seen it yet. Their scale-up strategy? They're Big Iron guys. When are they going to catch up to us on those fronts? I'm looking for their low-cost systems; I've missed those. The systems our partners are putting in the market are 30% or 40% of the cost of some boxes. When are they going to provide the ease of use you get when you use a Windows system as opposed to a Sun system? Oh sorry, I'm missing that. When are they going to give you the types of development tools you can get in the Windows platform for building applications and deploying them flexibly? I missed all that. I guess you could say there's a level at which we're playing catch-up on scalability and reliability, but they haven't even joined the battle on these other fronts. When's it coming, guys? These other things matter. I was talking to an ISV today who said 30% of their business is based on Windows and 70% on Unix, not all Sun, and Windows is growing like a weed as a percentage of their business.
Another of Sun's arguments is that its integrated hardware-software offering is more reliable than Microsoft's model, which relies so much on OEMs.
They're defending the model of the '70s, it's not the model of the '80s, '90s, and 2000. For any short period of time, the old model can have advantages. It had advantages for Apple for a while. But the model for this century is that specialisation and focus works. One of the things we're doing with Datacenter is we're making sure we're getting the best of both worlds. The best of the open hardware environment, the best of what that means in terms of price and flexibility and innovation. Plus, we've restricted a little bit; we have just a few partners and we're really doing the testing and quality control in a limited set of environments, so the customer can have his or her cake and eat it too.
How hard are you pushing customers to migrate from Sun Solaris?
There's two different things. One is getting customers to migrate, and two is getting customers to build their new applications on our platform.
Ballmer: I'd say most of our effort is the latter, not the former. Although Home Shopping Network and a couple other customers we have with us today will talk about migrating off of Sun and Oracle. But I can't say that's the low-hanging fruit. It's mostly about the new applications. Our pitch is pretty straightforward: You have all the "abilities" -- scalability, reliability, manageability -- and we give you the advantage of a tool set, a development approach, the embrace of XML, the things we're trying to accomplish in .NET, tied to the rich client. Despite all of the yicky-yacky that comes out of the mouths of guys like [Larry] Ellison and [Scott] McNealy, the world still uses rich clients. That's not to say every application is built using more than the browser, but the world is still built using rich clients. The browsers are getting richer, the intelligence in handheld devices and TV sets is getting richer. I just think Sun and Oracle are kind of missing the boat.
So migration is the harder sell.
Getting somebody to change something, there's got to be a pretty good reason. The world's moving too fast. We just bought Hotmail, and Hotmail was a big user of Sun and Unix, and we just recently finished the migration. You say, "[The purchase] was years ago." But we were trying to run a business, and we had more incentive to migrate. We migrated from Unix servers to Windows servers, we now can use half as many servers, and the cost per server has come down. Everything about that migration has been a huge win. Hotmail is the largest application site on the Internet.
How is Datacenter going to evolve over the next couple of years?
We're going to have the .NET frameworks. They will be integrated into the Whistler/Blackcomb timeframe. They'll be integrated into the Datacenter version so that infrastructure comes built-in. There are ways in which we want to improve the host ability of the .NET frameworks in that environment. That'll be a hosting environment that'll be very important. You'll see us make pretty much everything in the product operable in a headless fashion, with full scripted ability in the environment, which is important in the Datacenter world.
How do you expect customers to use 32-processor machines?
Big database applications would be the most typical. In some cases today you'll want to use Datacenter server consolidation, where people have a bunch of servers that are independent that some guys like to consolidate on one big honking machine. Server consolidation will be an area that's important for the 32-processor environment.
If Datacenter has the first .NET technologies, how big is the gap that lies between what you're shipping today and what the ultimate .NET vision is on the whiteboards in Redmond?
Well, we're not showing the user interface today. We do have a pre-beta out on the .NET frameworks, but we have more work to do. We need to make sure the servers are well-integrated with the .NET framework and we haven't even debuted -- except kind of, sort of with Passport -- the first of the .NET building blocks. We're scratching our way forward, but there's a lot to do.
What role will your partners play in .NET, besides building .NET devices?
They'll be building .NET applications, .NET devices, and we also see a world in which we'll be offering certain .NET services and people will be able to customize them, repackage them, and integrate them with their own services and make them a unified offering to the customer. I think that'll be one of the great wins, when you can hook up your SAP system and your Office applications and you have great XML [interoperability]. ... We're not there, we can encourage ISVs to do that with what we have on the market today, but there'll be a time a couple of years from now when we'll be able to do that. In the meantime, we have some things which are incrementally valuable. Visual Studio .NET will be the best environment for writing Web-based applications; these servers are the best environments for doing what they do, but they all fit in this kind of strategy