Charles Phillips knows the software market. Before moving to Oracle Corp. in May 2003 and working directly for Chairman and CEO Larry Ellison, Phillips analyzed enterprise software with Morgan Stanleys Institutional Securities Division. Phillips, an Oracle executive vice president, talked with eWEEK.com Reporter Matt Hicks about Oracles next big push—grid computing—and the companys future plans. Oracle during its Oracle World conference in San Francisco next week plans to unveil a new grid-based release, called 10G, covering its database, application server and enterprise management tool.
eWEEK.com: How does (the grid approach in the infrastructure products) affect whats happening on the applications side of the business?
Phillips: The applications naturally benefit, and the key reason is because you dont have to change applications to take advantage of it. Certainly ... the fact (that) as an applications vendor we can explain how to make the applications more reliable and more robust by running them on top of a grid gives us an added dimension just because we understand it well. Its something, actually, that other application companies can take advantage of. Its not exclusive to us, but its our knowledge of it and confidence in it and enthusiasm to use it. …Well be explaining right away how it relates to our applications business. …
An example that we like to give is that order entry is an application that is somewhat seasonal depending on the end of the quarter … So that application may become stressed and need more capacity. Something like HR, which is also seasonal but on a different time period when youre paying people once or twice a month, those servers may be idle, [and] we can load-balance and adjust for that so you dont have to roll in another server just for order entry because its the end of the quarter. ... [That] is a way that we can explain the benefits of the grid through our applications because were in both businesses. A business person may not understand all the underlying issues underneath the grid, but when you start talking in terms of an application, they know that problem, and this will solve that problem.
eWEEK.com: Is the whole 10G push leading to any significant changes to the applications as well, or are they really two different development tracks?
Phillips: We do have separate development organizations, but we always synchronize, (since), of course, our apps will be taking advantage of the grid. Obviously, theyre separate products.
eWEEK.com: Looking at customer side of this, what kind of response to the grid approach are you expecting from them? How are you going to convince them to upgrade at this point, given the economic situation were in?
Phillips: Save money. We have the problem I just described; [is] the answer now to go out and buy another server for order entry when you have idle capacity on the HR server? If I can tell you that theres no need to roll in a big new box, that you can redirect that application and take advantage of that excess capacity thats elsewhere in your blade farm, (then) you want to do that. We offer them the ability to grow incrementally and pay as they go. Instead of buying on demand, which is kind of the IBM approach, we allocate on demand.
eWEEK.com: Thats been a focus for Oracle for some time now. Little more than a year ago, there was the introduction of greater support for Linux. Is this part of a larger strategy?
Phillips: Id say its certainly a continuation of our embrace of commodity computing. We like low-cost components … Intel servers (and) Linux. We have the advantage now of having the semiconductor curve working in our favor as opposed to being on higher cost hardware where Microsoft was (the one) on the low-cost platform. So the tables having really turned where we are emphasizing the low-cost components and we tie together with software and a virtually free operating system... We scale with the low cost components and build what looks like a big server on our grid, not trying to do big SMP boxes to get the scaling. So our messages have basically flipped. They used to say to buy low cost components and PCs and tie them together, and now they dont. And now were saying it because we have this breakthrough in grid.
eWEEK.com: What drove the original look toward grid computing?
Phillips: This is basically 15 years of engineering history. To get to grid, you have to dial back the clock. First we had Parallel Server and learned how to build parallelism into applications, but back then it was manual and you had to change the application to take advantage of it. Then we said it would be nice if we could make that so you didnt have to change the application. And that took a couple more years and came out in the form of RAC.
Again, we said, "If we can make it so you can parallelize an application automatically without changing the application, what if we made that automatic when you continued on other nodes in a blade farm?" Call that the grid, where (you) discover those new processors and (its) automatic…You needed the 15 years of engineering history to get there. Its not something you do overnight. It was a progression of the natural thing to do: Make the application run unchanged and then make it automated to call it a grid.
eWEEK.com: But has there been recognition in the past few years that Oracle needed to expand who could really operate the underlying database and applications beyond large implementations and large machines? Was there a conscious effort to see how the technology and business direction could mesh?
Phillips: Thats been going on forever. We always want everybody to run on an Oracle database. At one point, we had over 100 platforms we were supporting trying to get to that very goal. Now there are fewer platforms relevant today, but theres not an important platform we dont run on. That pervasiveness of the Oracle database continues, and we think more people run Oracle than anything else so it seems to be working.
Discuss this in the eWeek forum.