When you think about Oracle, open source isn't the first thing that comes to mind. The company is best personified by Larry Ellison, Oracle's jet-flying, yacht-racing billionaire CEO (with a penchant for Japanese architecture). The open source movement is not supposed to be about enriching billionaires, and yet Oracle's Linux support, along with IBM's, is one of the best things to happen to the operating system, helping give it respectability among CIOs and IT staffs.

Oracle's history with Linux dates back to the late 1990s, a time when Oracle invested in Red Hat (along with IBM, Compaq, Intel, Netscape and Novell). Oracle then took aim at IBM with its Linux versions of the Oracle Internet Application Server 8i and Oracle E-Business Suite 11i. The company was unphased by SCO's 2003 intellectual lawsuit and still indemnifies customers against any copyright infringements, though the possibility of that happening now seems more remote. Today, not only is Oracle is still competing against IBM for share-of-mind among corporate Linux installations-but has taken on Red Hat itself in the lucrative area of Linux support. That move has not been enough to tempt some Red Hat Enterprise Level customers, who worry about incompatibilities. Oracle says they need not worry-and the company has so far resisted any temptation to come up with its own Linux distribution.

Oracle is also its own best example of a large corporate Linux installation, with some 10,000 Linux servers amassed at a Texas data center. Another Oracle Linux installation, Amazon.com, operates what the analyst firm WinterCorp calls the sixth-largest commercial database in the world-with a distributed front-end built on the open source Oracle Berkeley DB and the back-end built on the closed-source Oracle database. Other Oracle Linux installations include CERN, Sanyo Semiconductor, Electronic Arts, and Dell, To catch up on where Oracle is headed with Linux, I spoke with Monica Kumar, who leads the team responsible for Linux product marketing and business development. Coming to Oracle ten years ago from Informix, she first managed global partner alliances and worked in Oracle's Database Product Marketing and Server Technologies division. She holds a bachelor's degree in mechanical engineering and a master's degree in business administration.

How did Oracle and Linux first connect?

In 1998, Oracle 8 became the first commercial database to run on Linux-which, looking back, was a big milestone for the operating system. Until then people used Linux largely as a development platform-there weren't any real deployments at the enterprise level. With Oracle now behind Linux, that perception started to change.

Tell me something about the Oracle's Linux development team.

There are two different types of development. One group is involved with Oracle products running on Linux: not just Oracle Database 10g, which is now in release 2, but our Oracle Fusion Middleware and the full E-Business Suite of applications, which we moved to Linux a couple of years ago. All totaled, we have over 9,000 people at Oracle who use Linux on a daily basis to develop Oracle products.

The second category of developers is represented by the Linux Kernel Team, who've been in place for over five years now. This is not a well-known fact: people don't perceive Oracle as having such a strong focus on kernel development.

People have long speculated that Oracle would come up with its own distribution.

Our strategy has been not to fork Linux, but to enhance the core kernel. We've worked with the likes of Linus [Torvalds], [lead Linux developer] Andrew Morton and the community in general, in enhancing the core rather than creating something of our own.

There have certainly been plenty of forks.

There have been. But in our eyes, Linux is a commodity. Oracle doesn't gain any competitive advantage with our own version of Linux. Our advantage is in superior support delivery-in bringing enterprise class support to Linux. That's our focus. That means that any customer who calls us for issue resolution will get a patch, right away-even if we need to write code to resolve the problem. And of course, under the terms of open source, we make this patch available to the community under the GPL [GNU General Public License], thereby making it available to any distributor who wants to pick it up.

Where does Linux stand as an enterprise OS compared to its competitors?

From our experience over the last few years, Linux has made it onto the short list. It is one of the top three enterprise operating systems, which also includes Windows and the combined flavors of Unix. The selection depends on an organization and what their goals are. Some organizations want to adopt a commodity computing model and are looking very aggressively at moving everything to Linux. Some have a mix. Some even have Windows and Linux co-existing. But I would definitely say that Linux has made it into the mainstream.

So the squeamishness about using open source has been largely overcome?

To a large extent, yes, because Linux has been out for so long and because companies like Oracle have stood behind it. I cannot say the same for all other open source technologies. It's on a case-by-case basis.

What about lingering security concerns?

As far as I can tell, Linux has no security concerns that are Linux specific. Linux is now at EAL4 [Evaluation Assurance Level 4] security assurance level, in par with other operating systems. All of the methodologies that make operating systems more secure are now being applied to Linux. In addition, at Oracle, we do peer review of the code-with a second pair of eyes, apart from the developer going through the entire code to make sure it is secure.

Oracle talks a lot about grid computing in relationship to Linux. Do you assume that if you adopt one, you'll adopt the other?

For many of our customers, that's the case. That's because Linux is all about commodity hardware, and commodity computing is what grid computing is all about. Some customers begin with one and then consider the other: it goes both ways.

Is grid computing viewed as the primary way of adding computing power to a database management system?

Oracle certainly believes that grid computing is the way to go. The whole notion of clustering is that you can combine many small machines to get the power of the mainframe. That gives you much more flexibility, higher availability and all of the scalability that an organization needs today at a significantly lower cost.

Is Oracle now a Linux shop in its own enterprise applications?

Effectively, yes. Larry Ellison said back at 2003 at LinuxWorld that we would be running our business on Linux. We started moving toward that goal in 2003 and 2004, and today that's pretty much the case. Today, any new machine that Oracle brings in to our global IT environment is a Linux machine. We run all of our internal systems, employees systems on Linux. We run our on-demand business completely on Linux. We do our major product development using Linux. We have almost 10,000 servers in our Austin data center-all are Linux servers.

What's the story with "Unbreakable Linux?"

Unbreakable Linux is the overall umbrella program name for our Linux support program, which we announced last winter. It's enterprise-quality support at the same level and scale of support as we offer to the Oracle Database customers-though you don't have to be running any Oracle product. We set up the program because customers were telling us that their current Linux support was not adequate. We very closely track Red Hat Enterprise Linux distribution. We provide Linux software through Oracle's website if customers want to download it, which is 100 percent compatible to Red Hat Enterprise Linux.

What do you mean that existing Linux support is inadequate? How?

We had two key areas of feedback. Customers were really frustrated with the lack of the quality and timeliness of bug fixes. People told us that their Linux vendors took too long to make fixes or didn't make them at all. Or, maybe they did a one-off patch, but took too long to integrate it into a patch set or a new release. The second area was pricing-customers felt that the price they had to pay for Linux support was too high-especially given the whole notion of open source as a lower-cost alternative.

Some installations have voiced concern that Oracle support of Red Hat distributions would lead to incompatibilities.

The key here is that the Red Hat distribution is open source code. Other than our removing the Red Hat trademark, the only difference will be the few patches that we provide on a timely basis to installations that subscribe to Oracle support. Keep in mind that all these patches are going back to the community-anybody can pick them up. So we expect that at some point Red Hat might pick up some of these patches, as well, to include in their distribution.

I also want to explain how we synchronize with the RHEL releases, because that's a key point. Oracle makes every single patch that comes out of Red Hat available to our users. We do that within a few hours of when Red Hat releases a patch. And every time Red Hat comes up with a major update-typically every three to five months-we do a major synchronization. We take that source code, add any patches we've done in addition, and then compile the code. And of course when there's a major release, we do the same thing. So at every point, we are synchronizing with Red Hat. As I said, Oracle gains no competitive advantage to fork and end up with a different product. Our strategy is to stay compatible with Red Hat.

And one other point that is not well known: Oracle has been providing "priority 1" patches for Red Hat for the last four-and-a-half years-as a service to Oracle customers running on Linux. So any Oracle customer has been able to call us to report an issue with Linux running under a particular Oracle application-and we have given them a patch. And over this period, we have never broken any compatibility. The applications have been able to run as-is, and indeed, Oracle has ultimately provided the patch to Red Hat. That same process of patch delivery and maintaining compatibility remains ongoing today.

How are Oracle's Linux activities playing out in Japan?

In general, Linux has global momentum-and we've not noticed much difference by region. In the Asia-Pacific region, Linux adoption is very high and we see many installations looking to put the operating system in the data center. The only difference I see is the greater number of Linux distributions being used in Asia, versus the U.S. Here, a couple of flavors are dominant, whereas in Asia, we've seen more. I think that's why this whole Asianux consortium happened-as a means of providing a common core. Universally, I think, the idea is not to compete on product. Linux, by nature, is a commodity product-so why not compete on support and other real differences?

I noticed that you still play up the idea of user-indemnification against copyright infringement. Is that still a concern with Linux?

Somer users are still concerned-and want some form of indemnification as part of the support package. Oracle Unbreakable Linux support offers full indemnification to Oracle Linux support users against Linux intellectual property claims. A key reason Oracle had made this offer was to just eliminate this issue as a possible barrier to Linux adoption. We wanted to lay that one to rest.

Oracle seems to have an open source database product. What is the Oracle Berkeley DB?

It is a product we acquired through the acquisition of Sleepycat. It is an open source database, mostly for embedded applications. For Oracle, it is a way of filling out our database offering. Believe it or not, Oracle Berkeley DB has almost 200 million installations. It's shipped as part of mobile devices such as Motorola phones, in open source products, as part of OpenOffice. We offer it in a dual-licensing model: with a free version available for download and a commercial version that is available with support.

Given that many of your customers are now using an open source operating system, is Oracle concerned they may eventually migrate to an open source database?

No. We are actually seeing the market increase for Oracle products running on Linux, rather than the other way around. Last year, according to The Gartner Group, Oracle made $1.2 billion in revenues purely on databases running on Linux. The whole market is $1.5 billion-so Oracle has about 80 percent of that. The same goes for applications. We are seeing increasing adoption of applications running on Linux and other middleware products, as well.

Anything you want to add?

We've talked a lot about support, but there's also the whole notion about making Linux deployment at the enterprise level faster and easier. We're heavily invested there, as well.

A year ago, we announced a program called Validated Configuration, which is aimed at IT organizations that are interested in deploying Linux. The idea is that we do some of the configuration testing that usually falls to our customers. We've developed a test kit that not only tests Oracle running with Linux, but other components in the configuration-the servers, storage, networking components: basically, the whole stack. At the same time, we're working with our partners to put together various instances of the stack, running the test against them, and publishing the results. So the end user sees a best practices document that outlines what components we tested, the configuration parameters we used, the problems we encountered and the work-arounds we used. In other words, given this configuration, here is what you can expect.

The idea here is that we want to do more than just contribute to Linux as an open source OS. We are ensuring faster and easier deployment of Linux solutions, that these solutions perform and scale well, are highly available and secure, and can interact with other components in the data center.

Why have you taken this approach with Linux, in particular?

Because with Linux, there are so many possible configurations that customers were getting frustrated. Oracle heard this feedback loud and clear, and today we have published some 15 validated configurations with our partners-and we are producing more. All of our hardware partners-Dell, HP, IBM, Sun-are running the test in their labs.

What's the future for Linux in the data center? Will more companies look at grid computing as the best long-term strategy? Or will inertia prevail?

The future of Linux is bright. Many of our customers are investigating grid computing as the way to effectively use resources and modernize their data centers. But it depends how much the culture of a given organization accommodates change. Among our largest customers, we see both sides of the spectrum. Some are so leading edge that they want to try everything-even when it's leading edge. Others wait for three or four years, and even then, only deploy it on a limited basis.