Microsoft's Ward Ralston Details GUI-less Windows Server

Scott Fulton, BetaNews: System...Center...Configuration...Manager? The new titles always confuse me. One of the key features [of the replacement for Systems Management Server] is the ability to deploy installations remotely to thousands of workstations and servers. Of course, you need a public beta in order to see that happen. Somewhere along the line, some big company has to try this just so we can see it works. We need a Chrysler or an Alcoa to say, "We're going to rig Cleveland with Beta 3."

Ward Ralston, Senior Technical Product Manager, Windows Server Division, Microsoft: [On March 14], we kicked off our Rapid Deployment Program, so we have 400 partners that we brought on campus. We have some big names we're not ready to release, but you're absolutely right. These are the companies we are partnering with, both financially and with services, to create a partnership where we deploy our technology before RTM to get that exact type of evidence.

Scott Fulton: Yet you're in a rapid deployment program yourself, because you've got RTM in the second half of this year.

Ward Ralston: We already have 1000 servers at Microsoft already running Beta 3 of Windows Server. Interesting fact here, but we actually build Windows Server 2008 code onto Windows Server 2008 machines. So the actual ones that we compile on are running the 64-bit version of Windows Server 2008, which actually has halved our compile time overnight...And we can address incredible amounts of memory now. When we break that 4 GB limit on the x86 architecture.

Scott Fulton: You've broken it. It's time for the customers to break the limit, and that's going to be a tricky process.

Ward Ralston: Well, you're absolutely right. But our hardware manufacturers aren't even making 32-bit chips anymore. Every chip made is 64-bit, and they're dual-core, multicore, and that's only going to increase more. We've been letting our customers know that Windows Server 2008 will be the last 32-bit server that Microsoft makes, and that the R2 version that will come out in 2009 will be 64-bit only.

I don't think it's going to be that challenging in the server market as much as it is in the client market to make that shift to 64-bit, because the benefits are just so incredible: virtualization, WHEA [Windows Hardware Error Architecture], Terminal Services, when you can shatter that 4 GB barrier and have those additional registers to address.

Scott Fulton: Especially by virtue of deploying Office Communications server on top of that. The dream there appears to be to help convert Microsoft into a communications and computing company as well. It's getting into a space where it's not even the #3 player, knocking on the door of a market that has to this much been pretty much sealed.

Ward Ralston: Bill Gates summed it up pretty well when he said, we're seeing PBXes starting to go away, and data and applications are residing in the cloud. Of course, if you think about it, what is that cloud? It's a bunch of servers. So I think we are seeing this aggregation of all data residing ephemerally on servers, communication being one of them.

Scott Fulton: In getting that accomplished, moving toward a realm in which we have that much traffic, do we need to start rethinking the way in which logical networks are constructed? Right now, Active Directory networks are fairly hierarchically constructed. We're starting to see the need for logical design of domains that eliminates all that cross-forest trust and federation of trust. We'll start seeing clusters done a lot more heavily than before, and virtualization will completely change the game by giving all 64 processors a single face to the world. The tools are there to change server configuration overnight, to change the whole premise of how you design a network. When is it time, do you think, for system architects to go back to school and start over?

Ward Ralston: We're obviously not there yet. I think IT moves a little bit slower than radically. I don't think any organization's going to be open to radical change in their IT infrastructure. It takes a long time for them just to roll out Server 2003. I think we've given them the tools, though, to capitalize on logically changing their network and physically changing their network.

When you look at IPv6 and IPv4 running side-by-side, that breaks a lot of barriers with the IPv4 limitations in the namespace. Developing countries will be a huge user of IPv6. Traditionally, you've had the first computing countries have had a monopoly on the IPv4 namespace, and given a nod to Heterodyne, if you will, the other organizations' federated identities.

You mentioned cross-forest [trust], which was an incredibly arduous way to have an organization...

Scott Fulton: I trust you if you trust me...

Ward Ralston: Exactly, but it was even worse with NT 4.0 where you could create different trust models, but it ended up as a complete mess. Then we get tools in [Server] 2003. But one of the significant things we introduced in R2 was the notion of Active Directory Federation Services. Instead of having the federation have a forest level between two like systems, we can now take an authentication event in one organization and boil it down to its lowest common denominator - a sample token which can be consumed in a Web service and passed over into a WS-* compliant...[perhaps from] Oblix, it doesn't have to be Microsoft. That radical change is subtly happening with the tools and services offered now.

Scott Fulton: That same answer has been given for years, and the possibilities of network architecture have always been so far above what actually happens in the corporate scenario.

Ward Ralston: I couldn't agree with you more.

Scott Fulton: And the scope is only getting bigger.

Ward Ralston: And I think we give organizations the tools to capitalize on that, like MediaFS and given a nod to IPv6 too, but until there's a business need for organizations to have that fundamental architecture change, I don't think we're going to see it. But we definitely have the tools in place.