Lynn LeBlanc, CEO, FastScale

Lynn LeBlanc Sometimes, complexity begets complexity even when you try to make things simple.

IT managers for companies are either battle-testing or outright buying virtualization software from VMware, Microsoft  or some other provider to help combat the sprawl of physical servers in their datacenters. Virtualization software is touted for its ability to run several operating systems or applications on fewer machines.

This idea of using 25 machines to do the work of 100 is appealing to harried IT managers.

But virtualization, the industry is quickly learning, adds its own layer of complexity: Though there might be only 25 physical servers, there may be 100 operating system images an IT manager must keep track of from one computer console. This can get messy.

FastScale Technology, a startup founded in January 2006 by CEO Lynn LeBlanc, CTO Stevan Vlaovic and vice president of engineering Richard Offer, launched FastScale Composer Suite to help IT admins manage virtual servers in datacenters and Web farms.

The software, intended as a complement to server virtualization software from VMware, Microsoft and others, automates the building, managing and provisioning of virtual server software so IT administrators aren’t stuck trying to rein in various operating systems, update packs, libraries and scripts.

FastScale comes to market at a time when analysts predict virtual server sprawl will create a submarket for new software providers. LeBlanc discussed this emerging market and her company’s inaugural product in a recent briefing with

Q: What are the datacenter issues FastScale Technology is trying to address?

There is so much complexity and systems admin time spent building production-grade software environments. Part of it is the deployment and operation, but the really big part, and the area we believe where there has been limited innovation, is in the automation of building that working software environment on the front end.

Without doing that, a lot of consolidation falls on deaf ears because if you’ve got fragile and complex software, it has to be isolated from one another. When you combine that with peak load requirements, you really don’t have a choice but to spread it out over physical and virtual servers.

Our view is this root cause of bloated, complicated and fragile software makes this problem chronic. Most of the solutions that have been deployed in the last three to five years have focused on the back-end deployment and management but there’s been nothing on the front-end complexity.

Q: Are you saying the virtualization software from VMware and Microsoft is not working or causing more complexity?

I’m saying that they address a certain class of problem, which is once you build a production-grade environment, as long as your application isn’t too I/O intensive or CPU-bound, you can get some hardware gains by combining things on a single physical machine. But what they don’t address is the efficiency of what goes into those virtual containers.

We’re starting to see more worry about virtual server sprawl just like we saw physical server sprawl, because, while it may be easy to crank out a container, it’s just as hard to manage them as it is to manage thousands of physical servers. We work quite a lot with VMware and their ESX Server. By creating a very efficient software environment that goes into the ESX container, customers can run five times as many containers with the same level of performance.

Each container consumes less resources if you’ve got a more efficient software stack to start with. Whether you’re using Opsware to provision or VMware to consolidate, our focus is to address the root cause of software complexity.

Q: This means running FastScale’s Composer Suite in conjunction with VMware or Opsware products. Aren’t you adding to the complexity by adding another piece of software for an admin to worry about?

No, we attack fundamental problems. FastScale Composer Suite is non-invasive to existing infrastructure, operating systems and applications. In a traditional environment, whether you have a single server or golden image, you have one or more applications — an operating system that supports hundreds of thousands of applications, so it has quite a lot of capability that your application is never going to exercise. From Apache to IBM WebSphere, these applications call just a very small subset of an operating system.

There are three core technologies. First, our Application Blueprint technology is a series of algorithms that enable us to determine what part of a full operating system will be called by that application. We do this automatically, in seconds and at the application’s runtime.

Second, once we’ve identified the smaller set of resources, we can build a very small self-contained package of the application resources, or the Dynamic Application Bundle. What that delivers is about a 99 percent smaller software environment.

Third because the environment is so small and so lightweight, we’re able to provision the package to bare metal in seconds instead of an hour. The application software stack is small enough to run in memory, and this non-persistence gives customers flexibility to move workloads around.

For many customers, the ability to run in a diskless mode is a big deal because it reduces hardware failure and power consumption. When a job is complete, the server can provision a new application bundle in less than a minute.

Q: When you look at the competitive landscape, who do you see trying to solve the same problem?

I would say our biggest competition right now is all of the manual process that large organizations do to streamline their software stacks. They’re proprietary tools that people build internally to streamline one applications at a time.

By keeping it to basics on the biggest root cause pain point, that gives us the best opportunity to set a standard in how complex software gets built, managed and deployed.

News Around the Web