Suns Server Chief: Systems Road Map Shows the Way

Q&A: The latest upheaval at the OEM will not impact timelines for Opteron-based or SPARC servers, according to Sun Executive Vice President John Fowler.

Sun Microsystems is a company in a state of flux, undergoing an aggressive restructuring while continuing its rollout of new products, as the July 11 launch of three new Opteron-based servers—including a blade server and scalable rack-mount system—showed. Among the key changes was the combining of Suns Opteron "Galaxy" systems unit with its traditional SPARC business. John Fowler, executive vice president of the new Sun Systems Group, spoke with eWEEK Senior Editor Jeffrey Burt about the new business unit, Opteron and SPARC servers, and how the Santa Clara, Calif., company is working with its customers during such a time of dramatic change.

Whats the advantage for Sun and for your customers in combining the two server groups?

We have been sharing engineering activities in the past between the x64 and SPARC teams. It obviously makes it easier if youre one organization.

The more interesting thing comes around in other parts of the organization. For example, operations, supply chain and marketing. Theres a whole bunch of areas there where we can position and market the products as a holistic family regardless of what chip is included. Certainly well do that.

Then on the operations side, as we share more engineering things, theres a bunch of things we can make more efficient there as well.

There certainly seem to be benefits for Sun. Where are the benefits for the customers?

What youre going to see is a lot more work actually to create a product line that people can deploy in multiple architectures. What you saw with Galaxy and Niagara systems is a shared mechanical design. What youre going to see going forward is a lot of sharing on systems management and technology on the systems themselves. And obviously, with blade products, youre going to see blades that include both SPARC and x64. Putting the groups together just makes it easier to do all of those things as a cohesive team.

Again, one of the things I said before is this is something we had actually planned for a long time. We first wanted to grow the x64 business to be pretty healthy for us, then look to combine the organizations. So were actually following through with something we planned for a while.

Regarding blades, youve got the Opteron blade on deck, and Opteron and SPARC blades for telecommunications companies. Do you expect to roll out SPARC blades for the general customer?

We havent said specifically what kinds of blades well be doing, but we are going to be—and we announced on [July 11]—that in the future, blades platforms will include SPARC. Its just that the initial announcements happen to be x64. I havent said exactly what SPARC [chips] theyll contain, but we certainly will have SPARC blades.

The reorganization kicked off a lot of speculation regarding the SPARC side of the house, speculation that the cuts were going to impact that side in particular. What can you say about that, and will this mean a greater reliance on Fujitsu in future development of SPARC?

Obviously were also looking to tweak the expense profile of the company, and certainly Sun is doing it now. What I will say is there really are no changes to any of the publicly stated systems road maps at this point. Were continuing forward with the APL [SPARC-based Advanced Product Line] with Fujitsu as we previously worked on the high-end machines, and we still have the very robust SPARC portfolio with "Niagara II" and with "Rock" coming, and all the systems that are around it.

/zimages/2/28571.gifIs Sun ready to go up against IBM and Hewlett-Packard in the blade server arena? Click here to read more.

I think the thing you will see is a lot more common mechanical designs and common systems management, [and] other parts of the systems that are shared more and more between x64 and SPARC, for kind of obvious reasons. Not just because it makes it more efficient from a [research and development] standpoint, but it also makes it simpler for the customers.

Regarding the shared infrastructure, what will all this mean for customers two, three or four years down the road?

What people will see is the ability to plan a data center for SPARC as well as for x64 using blades in racks and using a set of common thermal and mechanical designs across all of those, so you only have to plan them once.

But more interestingly, [customers will see] a very common systems management. For example, on the x64 side, you see technologies like IPMI [Intelligent Platform Management Interface standard] and others that are really considered more x64-oriented. Youll see technologies like that also coming to the SPARC program, so, as a customer, once you get used a certain set of systems management [capabilities], youll be able to use it all across the infrastructure. Thats going to be easier to deal with.

Does the fact that Sun has put the person who was in charge of the Opteron-based systems in charge of all systems now give any sort of signal to the industry regarding the importance of SPARC going forward?

Ive only been the Opteron guy for the last couple of years. Ive been the Sun guy for 16 years, so people whove been around Sun for a while actually dont think of me necessarily as the Opteron guy. They more commonly think of me as the software guy, because Ive done that for a much longer period of time than for Opteron.

I think people inside Sun dont see it as that. They just see it as a logical evolution of the platform. On the outside of Sun, theres been really no question in that regard, and I think the reason for that is we have such a portfolio of products. We just came out with Niagara, weve talked about Niagara II being not too far away, and we have Rock under development, and … UltraSPARC IV+ is not very old, we have APL. So we have a tremendous amount of products on the SPARC side. I think thats what most people look at to decide if something is happening or not.

As I work with the teams, its pretty obvious: With SPARC we can innovate at the microprocessor level, as well as all the systems designs and software. With x64 we can innovate with systems designs and software throughout. Were not actually designing the microprocessor itself. Theres still just a lot we can do on both fronts.

Next Page: Customer confidence, the Opteron business and Intel chips.