Bill Cook, a 19-year veteran of Sun Microsystems, recently decided to explore the challenges, risks and rewards...
found in the start up environment as the new senior vice president of sales at San Francisco, Calif.-based Penguin Computing.
Cook said he sees "quantum leaps" forward for Linux clusters in the future as more enterprise level players adopt traditionally HPC-friendly technology -- but can system administrators get passed the "complexity" of such systems? Cook thinks they can and bets that his new company is perched on the cusp of a very big -- and very lucrative – Linux clusters wave.
After his first week on the job, Cook spoke with SearchEnterpriseLinux.com about his new role at Penguin, Linux clusters, and the challenges of bringing HPC to the enterprise.
What attracted you, a former Sun exec, to a start up Linux cluster company
Bill Cook: I started at Sun in '86, and then I kind of grew with the company with increasing more responsible jobs from the sales leadership perspective. When changes were made at Sun in January, they had asked me to do a different role. At that time I took kind of a pause and said to myself, 'after 19 years is this what you want to keep doing?' It was a personal decision, a decision where I really wanted to get with a small, growing company and take advantage of the experience I had gained over the years.
This is not an indictment of Sun, but they are a big company now, and I had decided to give it a shot at smaller company. Now specifically why I chose Penguin Computing, were that they were in a growth market, Linux, and had some revenue and a strong customer reputation in their market as a good company to deal with. And then they also had an intellectual property engine built around Scyld. Still fairly small company, but they have an aggressive growth target.
What has the adjustment been like, moving from a company so large, to one with 80 employees
Cook: Well, you're asking me five days into it, so ask me again in a few weeks or months. In running a big organization at Sun, in some sense people think it was huge responsibility, which it was, but when going to a smaller company, you feel an equal level of responsibility, in that things are more personal. Even though Sun became this big company with a huge staff, it never had that kind of hierarchy view of empowerment. If you know Scott [McNealy], that's just not him. He tried to keep it pretty lean. The company is smaller certainly with a different feel, but in some sense you are involved in broader issues here, more market and product activities.
What is your number one priority now that you're with Penguin?
Cook: Hopefully it's pretty obvious – to generate revenue; strengthen Penguin's presence in market; Scyld clustering; and to build the customer installed base.
What are some of the challenges you expect to face as you market Linux clusters to market? How deal with them, what points stress to eliminate misconceptions?
Cook: Think the challenge is always, you know, to turn complex issues of HPC into more simplistic policy. On the product side there is a lot of complex technology under the covers in Scyld; all the moving pieces that make clusters work. One value proposition here is that we make it simpler to install procurement and administer so you can stay focused on the job at hand.
Are enterprise administrators intimidated by HPC technology?
Cook: I don't know if I would say intimidated, but I think there is a need that they all understand; that going towards Linux type systems is a more cost effective in terms of hardware with just blades or entire systems. There's an understanding that there is going to be a need to cluster these systems together to take advantage of their power.
However, as you move out of HPC world to enterprise you are increasingly seeing questions that there will be bigger challenges. The HPC world is more comfortable with complex systems, and quite frankly that is part of comp advantage of Penguin, one that we drive home with Scyld -- it makes it not intimidating.
The market will dictate how quickly occur, as we for better and better job of that market account ... Even the most fearful system administrator will become at ease with the technology as you demonstrate you can install and run these thing. No one is arguing with the clustering idea, and I think this point in time is a stepping stone to make sure people are comfortable with the idea of applications running in any environment with a system load in applications place, and clusters are a nice step in that direction. Quite frankly customers are more comfortable with clusters than they are with virtual grids at this point.
Who is deploying clusters today? Who would you like to see deploying them more often
Cook: You are still seeing it in the academic and research spaces, the traditional HPC places. As far as enterprise level clusters go it is a matter of acceptance from an applications perspective. I would say today that enterprise HPC is in a type of early adoption phase. Quite frankly this is more akin to when Sun started as well. There have been quantum leaps forward here and there's no end of these trends in future.
What is the competition like in enterprise HPC technology?
Cook: Hard to pin point a specific competitor, as there are a lot of different, rages of cluster offering companies from open source vendors to more traditional players from the Unix space. [The market today] is more of a coalescence of technologies. I think it is more race is on form dif angles, hardware centric view, software centric view, blend of this things… that's why I said Penguin in a unique sweet spot on the convergence of those things. Eventually as the market starts to mature you will start to see two or three players become major operators in a given space, who will then battle in out and we'll end up with few. But right now I think it's too early to call.