Bill Coleman, CEO, Cassatt
Page 1 of 1
Bill Coleman has some experience starting companies. After all, he is the "B" in BEA Systems, a middleware maker who made a killing in the application server space throughout the 90s.
During Coleman's tenure as BEA's first CEO, the company became the fastest software firm ever to top $1 billion in annual revenue.
Now he's trying his hand at what he believes will be the next major wave of computing. His San Jose, Calif., startup, Cassatt Corp., is looking to solve data center clutter by consolidating multiple machines onto one with software.
Coleman told internetnews.com that his company is in the "calibration stage" of targeting the right customers with the right solutions, and is in sales engagements with over 20 major customers, including Pfizer, Cisco Systems and the CIA.
Q: What value does Cassatt offer to people at a time when virtualization technologies are gaining momentum?
What we do is automate the ability to guarantee delivery of service levels from an IT infrastructure. The closest description of the way we operate is what Gartner calls real time infrastructure. We turn the entire computing infrastructure, hardware and software, into a utility. Then we use a policy-management framework to generate services of that utility on demand.
Q: The description sounds a little like what Opsware is doing.
Opsware automates system management. They do not automate the runtime. We automate the runtime. We virtualize the application space and use the physical infrastructure as something we can redeploy on the fly.
We don't virtualize the hardware or the storage and we don't do system management. There's three aspects to it: the virtualization of the software from the Hardware; the setting of the policies; and then the dynamic management of the system.
We will automatically discover everything that's out there, profile it and, based on policies set by the customer and the demands of the application and its workflows, we will meter that capacity and change the profile by the priorities and policies on the fly.
If you want to set a policy that for the last day that should run and guarantee 12 seconds throughput for any order, we'll take all available resources and harvest systems to deploy them on the fly.
We not only virtualize the applications running on the operating systems. We can also do this for Web services running on Web application servers and we can do them in combination, so it makes it a true workflow environment. Today, we support Windows, Linux and Sun Microsystems' Solaris.
Q: What are you doing to build on what you already have in Web services?
We'll be announcing the Web Automation Module as an option to our product. It allows you to define the policies on how Web services interact. So, Web services can run on one or more Web application servers, independent of what they're running on. We can actually automate cluster management, which BEA and Oracle can't with their database or application servers.
We can also set the policies on how the Web services themselves are run -- what kind of utilization and business interactions they take.
Q: What are the differences between what you and other virtualization providers offer?
We do it in a scale-independent way, which no form of physical virtualization can claim. In other words, the amount of overhead for managing one server and 1,000 servers is the same, whereas with any form of system management or grid computing or virtualizing the network, the overhead escalates with the amount you're doing, and that's the problem with scale-dependent.
The secret sauce of doing it that way is that we don't put one line of code on any server we're managing. It's all done through our console and the standard interfaces and watchdog timers that are available with the operating systems.
Q: Who else do you bump up against in this category of real-time infrastructure from Gartner? Would you list the IBMs and Suns as providers of this sort of technology?
This is an emerging space for Gartner. They're claiming that this is what they believe is the future of the data center. They characterize IBM's utility computing product -- Tivoli Intelligent Orchestrator -- and Sun's N1 as sort of the two potential candidates.
The issue is Sun hasn't really shipped an N1 and Tivoli Orchestrator is very early. Gartner has no reference cases yet. Then Gartner characterizes a number of different start-ups, all playing in different parts of the space. Certainly us, and a few others. But they think it's an emerging space.
Q: Do you feel startups can be more agile to get ahead of these larger companies?
I absolutely do. The invention of something new takes someone who is totally focused on it. It's their only business and their survival. If you really look back for the last 30 years, every new level of the platform ended up being defined by a startup.
Some of those became really big. Like Intel, Microsoft, Cisco, EMC, Oracle, etc. Some didn't make it, like Netscape, Compaq and Veritas, etc.
If we're really successful, the hardware guys sell a lot less hardware and a lot less services, so it makes it hard for them to jump ahead of that curve.
And the reason why customers want to buy it is to buy a lot less hardware and a lot less services. But the way these things always go, you have three to five years during the definition phase in which one or two startups will emerge and start growing as they define the platform.
Once the technology settles down, the big guys get to be good enough and you have to figure out if you can move beyond that space or end up like Veritas or Netscape.
Q: How has your experience of starting something like BEA prepared you for this?
I know a lot more what to expect. It's also taught me to be a lot more concerned about the quality and experience of the people I bring together here.
For a software company, there are two first-order variables that can put it together: the market opportunity and the team. If you have the right team and the right market opportunity, everything else will take care of itself.
Q: What are your short-term goals for Cassatt then in terms of the market?
We're getting our first deployments out and gaining "referenceability." My goal is to have five to 10 very visible, very credible accounts by mid-to-late spring and to calibrate the sales model to determine how fast it would be effective for us in scaling sales.