BizAnalyst Network

Question: What are the concerns and questions of new clients to SpikeSource? What are they looking for?
Ruiz: Generally, the concerns fall around the ability to support and to meet SLAs for the offerings that we have using open source. Especially with some of the larger companies, who have been early adopters, specifically in financial services, what open source allows them to do is not just to mitigate or reduce their license costs but it allows them to rapidly deploy a solution that directly affects their end customers. So that shrinks time to market. Then they find out they have to build their own internal support mechanism in order to maintain the infrastructure. From there, you have to have either expertise directly in the components that you've deployed, not just from a developer perspective but from a support perspective, and then you need to interact with many different constituents, like the people who actually own the source code. All this is aggravated by the fact that the open source that they've put into play isn't just one unit; it's many. All these components are on an asynchronous release rate. So if you've got a patch that affects security in Apache, that may influence or directly impact the database, the indexing engine, the application, the operating system; the repercussions of that can go up and down the stack. So it's not just about being able to source a patch for that component, it's about being able to test that across the board.

Question: Is the demand stronger for the defined, pre-built configurations, or is it spread between those and the option of the customized, made-to-order stack?
Ruiz: I think this goes to the credit of the engineers who have originally put the stacks together, but the demand is slightly skewed toward the pre-built. The largest customers want slight or major tweaks, but if you just look at raw numbers, that'll go to the pre-built. The pre-built stuff is basically getting the developers to a high enough watermark where they can start adding value. Here's a LAMP stack or a LAMPJ stack, here's a Web services stack; from there, they can start adding value. So they're very common building blocks.

Question: How is the configuration manager affected by the choice of pre-built or made-to-order configurations?
Ruiz: From a technical perspective, we've broken up the configuration manager into a couple of toolsets. ... Whether it's a pre-built or custom stack, there are only so many permutations of the custom stack. We have in fact built a stack integration module (SIM). Any permutation of the stack that you could do for the made-to-order stack is already taken into account. So we have a configuration file for any permutation. The reason we've done that is because we needed to test all those permutations in order to ensure we're delivering a valid configuration to the customer. That's part of the beauty and strength of SpikeSource; the technology here is extremely strong, not just from the point of view of a nice, well-integrated stack, but because we have this large, automated test harness. We're able to take stack combinations and run them through the test harness automatically. Then we're able to use our SIM to come up with configurations that are suited for each of the permutations. It's not just about supporting the stack, it's about the whole process of being able to automatically test, configure, deploy and update it. It's pretty sophisticated in the background.