How would you like a silver bullet you can drop into your data center and have it take control of business-service flows and reduce costs, all without requiring you to change hardware, software or your systems-management and production controls?

That's the elevator pitch of William Coleman, founder and CEO of Cassatt, a 3-year-old company with a utility computing vision built on virtualization and automation technology.

Coleman, who also founded BEA Systems, starts with the assumption that IT has two core goals: 1.) regulate application service flows, an example being the processing of a transaction where failure of any part sabotages the whole thing; and 2.) do so in an efficient and cost-effective manner.

Companies are trying to address core application requirements today, Coleman says, using a variety of methods, including systems management tools and grid technologies, to scale environments as loads change. Although valuable, neither addresses the service flow issue, he says. And while virtualization is good for local optimization, it is hard to scale across the network.

When Cassatt was crafting its answer, it decided it didn't want to have code on servers, because of the inherent scalability limitations, and it wanted to deliver the product as an appliance to ease installation.

The company's software is delivered on a Red Hat Linux-based dual processor box that harvests run-time images of your servers. Data that associates those images with particular devices, such as media access control addresses, is stripped out and the images then are stored.

"We capture everything and then set policies about what hardware the application needs, the capacities, and how it scales out in a domain, meaning how it relies on and relates to other applications," Coleman says.

The controller figures out what is alive and what is running what by listening in on system-management streams, and it can recognize failures and reboot or add more capacity from a pool of spares. Servers change their colors as needed by PXE booting from the image library.

"All of this is self-configuring, self-optimizing, self-healing," Coleman says.

Service-flow control has to take into account virtual machines and Web services running above that, but Coleman says his silver bullet can change the profile of a workflow on the fly.