What specific changes have been made to the kernel that is leading to the stability and performance enhancements?
What we did was, we took certain workloads that we used our 2.4 kernel for, and we did longtime benchmarking and tweaking with memory management with the way hard disk and network reboots would be handled inside the kernel. To give you an example, in the year 2000 we had a huge customer with a cluster based on Alpha. We gave them a 25% performance boost by changing the No. 4 to a No. 8 somewhere in the kernel. This realignment of bytes within the memory made that much of a difference. These are the things you have to do. It's something that simple, but it took us three months to discover that. I don't have examples as spectacular as that for the 2.6 kernel, but it shows you I like that example because it shows how the work has to be done and is being done. How important is it internally to be the first out with 2.6-based technology and beat Red Hat with it?
Actually, when we started to lay down our road map, we hoped it might coincide with the [2.6] kernel release. But that was not the primary goal. It is a very nice thing that we are going to be the first enterprise-class Linux out there, but predictions about kernel release dates are like trying to win the lottery. It was clear it would happen during the last quarter of 2003, or the first quarter of 2004. But if it were the end of March, there would be no way for us to have it stable. Right now, it played along quite well and fit perfectly with our road map. How much enterprise interest are you hearing regarding a 2.6 kernel version of SuSE Linux Enterprise Server?
You'd be surprised how much interest we are hearing. We announced last year that the next version of SuSE Linux Enterprise Server will be based on kernel 2.6. The big enterprises all care about two things: stability and performance. The 2.6 kernel is a vast improvement in both.
The thing is, we did some benchmarking; our highly tuned 2.4 kernel of the latest service pack, which is really fast, and high performance is as good as the untuned current 2.6. Now imagine what would happen if you really tuned 2.6.
We started 2.6 tuning when we did the last service pack in November, and we saw tremendous performance improvements. And, of course, when you look at stability, when 2.4 was released in 2000, it was not yet that clear what needs enterprise customers would have. So now we are one step further down the road. Now, everyone understands what the internal architecture of the kernel needs to look like.
There is a huge demand for it. They can't get their hands on it fast enough -- on a 2.6-based technology that we could offer. We have a technology preview that we put on for customers and partners; there was a lot of interest. What do you think was the turning point for Linux? Was it a technological development? A mindset change?
In no specific order: If you look at the timeline of the adoption of new technology in the enterprise, it takes them one year, two years, three years until they give up some real loyalty. In 2001, 2002, there was some real hype about Linux, and people started to try it out and figure out they can use that. That led to adoption.
Now, there are more and more applications coming, and you can do some real-world solution stacks to help customers run their business. There is a concentration right now in the Linux market on two vendors. This definitely helped as well, because the customers didn't have to fear what happens if the market moves in another direction and I'm stuck with this platform -- ISVs especially hate this thought. This would mean re-certification and re-porting. This helps people make up their minds. Then there are, of course, huge public events like the city of Munich deal that [was discussed] all over the world and made people think, 'Wow, Microsoft can actually lose.'
There are many points. In Autumn 1998, when Oracle decided to bring their offering to Linux. This was really the first milestone. There is a company out there that believes in Linux. That and, of course, the IBM investment in 2000.
These are the events that made it possible for customers to evaluate whether they can use Linux or not. Without that, the customers would be losing choices. There would be 50% of Web servers running Linux, but that would be the only thing.
I would point out another thing in the last year is that there are enterprise Linux offerings out on the operating system. We established SuSE Enterprise Linux Server, for example. This was a proven concept with maintenance, with certifications, with a huge band of hardware and software certifications that help companies make a strategic decision that makes them believe they can run their businesses on Linux. What kinds of questions or concerns do customers have regarding the acquisition?
In all cases, there was just two questions: Are all of our agreements valid? Is Novell continuing with Linux? In both cases, both answers were 'yes.' Every big customer that we have knows Novell and knows Novell is a very stable, reliable partner. When you talk to enterprise customers today, what are the conversations like today as compared to a year or two ago? I imagine you're talking to a more informed audience today.
You don't have to teach everyone what open source is all about. That wipes out the first two meetings with every customer. People have made up their minds that Linux is a viable OS. It is now an accepted fact that you can run your business on Linux. You don't have to talk people into trying out Linux, but you talk to people about how the technology of Linux can benefit their business. Now that Novell has announced its indemnification program, another whole set of meetings with the customer's lawyers has been eliminated. That eliminates that part of the discussion. Now that SuSE has the Novell brand working for it, what specific issues does this eliminate for SuSE with new customers?
If you are talking to an international bank and try to talk them into relying with the core of their enterprise on a small Germany-based privately held company -- you can figure out what happens. This is the biggest issue.
Now we have a far better reach. We have support centers all over the world. We have 3,000 people in the field who can work with customers, help build customer solutions, solve problems. This is an order of magnitude different than what SuSE was before. It's an order of magnitude bigger than Red Hat. Was there any internal apprehension about being acquired by a company that deals in proprietary software?
No. Novell has made it so crystal clear that they see the future very strongly focused around Linux. How is SuSE's integration with Novell going? What kinds of changes will customers be seeing?
We now have a chance to offer to the market a really unique solution stack because we can now start at the very bottom of the operating system, offer some layers in between, like some services and middleware components on top of the operating system, and go up in some cases where Novell offers applications like its collaboration suite. The is the first time with Linux where a vendor has a chance to integrate the components right off. This doesn't mean they will be hard locked together. They are still components but, nevertheless, as we can work on all areas of that stack, we can make sure it fits very well.
After that, the integration is going smoothly. The Novell people are really anxious to get moving on Linux, and they are really positive about us joining the company. SuSE people see the difference it makes if you are part of a $1 billion global company. That really makes a difference if you call on customers; there are no more questions about us being a small company. Those instantly disappeared. This is a huge deal for SuSE and for Linux?
This is a significant step. It's the first time that an open source operating system has achieved that level of EAL certification. For SuSE, it was a big step as well. In August 2003, we announced EAL2 certification, which was the first time ever that any open source piece that had received any type of Common Criteria certification, which means it is accepted in many countries.
This is also the first time that an operating system was certified on more than one platform. That has never been done [SuSE Linux Enterprise Server 8 was certified on all IBM hardware platforms]. That adds a whole new level of complexity to the task of getting the certification. What does SuSE Linux Enterprise Server 8's recent attainment of Common Criteria EAL3 certification mean for enterprises in terms of their buying decisions?
First of all, the EAL certification gives an assurance to enterprises that a certain level of testing has been performed, and it verifies that this product adheres to a certain protection profile.