In this Buzz on IT Automation Clyde Logue and Jody Hunt talk about something every IT shop contronts: dealing with bottleneck between development and operations in the release of new applications -- or as you’ll hear us say, bridging the DevOps gap. What’s particularly interesting is BMC has recently acquired a company that brings a tool to make a difference here.
Clyde Logue is Founder and VP of Product Management at StreamStep, Inc. and Jody Hunt is Lead Solution Manager at BMC Software.
There’s been a burst of folks talking about monitoring and metrics and DevOps. It must be the Zeitgeist because it’s been the dominant theme in my thoughts lately. Monitoring is core to DevOps for the systemic feedback loop (there’s probably two loops: one long which goes into requirements for the app and one short that is remediating problems in production). But I couldn't get my arms around why monitoring seemed intrinsic at the gut level but not intellectually.
I think the fellow who coined the term DevOps has just given us a key metric/dynamic to measure and, also, because we now know what we want to measure, how we can go about it. (Kudos to Jody Hunt for the succinctness of what follows.)
DevOps increases MTTF or, in other words, the duration of correct app behavior in production. If we use Patrick’s math from the preso, we get:
As MTTF goes up, it dominates the equation and the equation approaches 1 (i.e., you approach 100% Availability). There are two ways you can influence this rate of change, one is by decreasing MTTDetect and decreasing MTTRepair. Monitoring decreases MTTD while automation tooling decreases MTTR.
Patrick has, I believe, given us a core equation for DevOps. So, we now know how to answer a fundamental DevOps question: how do you know your application release / deployment is going well? Monitor it.