IDGE: I want to explore a little bit more about this, what you bring to the market. Obviously, where the software itself is open source, the whole value proposition is around what you deliver to make it easier for the enterprise to deploy.
Whitehurst: Let me spend a minute on that, because I think a lot of companies, even customers of ours, immediately jump to the thought that Red Hat offers support and services on top of free software. That is actually not what we do, or it is a very small part of what we do.
If you look at Red Hat, we have 93 percent gross margin on our subscriptions. In other words, support is seven percent of our cost structure. That looks very similar to any other software company. R&D as a percentage of revenue is 18 percent for Red Hat. It's 16 percent for Citrix, it's 12 percent for Oracle, it's 14 percent for Symantec. So even for being open source, we're spending more as a percentage of revenue on R&D than they are. The obvious question is: Why? Yes, we're getting the code elsewhere. But the real cost is productization. We take that [open source] code and we turn it into a product and commit to supporting it for 10 years, there is massive cost associated with that.
Let me use a simple example. Let's say you're the New York Stock Exchange and you implement your trading platform -- they're a big reference customer of ours -- on Red Hat Enterprise Linux. You probably spent a lot of money to implement that, and you probably want to run it a long time. Well, three years from now, Intel comes out with a new chip, and you want to use that new chip because it will make your trading platform faster. Well, the beauty of open source, Intel enables that new chip in the new kernel. But if you're using a three-year-old operating system, you don't have to completely port your application every time there's a new piece of hardware that comes out.
Red Hat maintains each of those kernels for 10 years. We put something like 150 people against each release that have to stay with that release for 10 years. When someone fixes a bug upstream in the new kernel, we have to track that and backport it and patch it into all the old kernels. When Intel enables a new chip or anybody enables a new piece of hardware, we have to backport that into those old kernels. So we have a massive work structure to maintain those to provide that stability.
Open source doesn't do that. Open source is a development model. It's not a productization system. We do that. And so our cost structure actually looks a lot like a traditional company because all of that's really expensive. We are not a services and support company, we actually are a software company, and we sell a subscription to that software.
People may think: Well, I'm just going to download and use some random Linux. But what are you going to do in two years when a new version of that comes out and your application is not going to run? That's what we do, and that's our value. If you simplify it in very high-level terms, we take the power of the open source production system, which has a lot of benefits, not only Intel enabling their own hardware, but people like Google helping draw the road maps to meet their needs. So you can feel confident and sleep well at night that it will probably meet your needs in terms of scale. But we make that consumable for an enterprise in a traditional way.
And again, yes, we provide service and support, but that's a tiny, tiny piece of our overall activities and a small minority of our cost structure. Our cost structure is driven by taking those things and making them products. At any given time, we have something like 72 kernels that we have to support, because we support these things for 10 years. We'll come out with a new version every two years, and then you get different flavors. It's a massive engineering organization to support and drive those products. And that's just Linux. Obviously the same thing happens in something like JBoss.