I recently heard an industry analyst state that IT is locally brilliant but globally stupid, while referring to the release management process. He drew an analogy and asked the question: are we rewarding the fireman or the arsonist?
Interesting question. Many of our customers report that application-centric deployments into pre-production and production environments are largely manual and script driven, with much dependence and reliance on the few resources currently tasked with defining and executing deployment tasks and activities. There comes a point with deployment when rewarding the fireman becomes akin to rewarding the arsonist, only they’re one and the same and the only resource who knows it, sees it, and can do something about it.
In a recent separate call with a customer, I met with two release analysts, a DBA, a resource responsible for middleware, and the release and production control manager. We learnt of 12 application types, 8 of which are web-based applications using WebLogic and Tomcat, and a backend Oracle SQL database. The customer reported an increase in application deployment requests to 60 per month, and has an average application deployment time of 38 minutes. The release analysts spend all their time creating and maintaining deployment scripts, managing the coordination of manual tasks and activities with DBAs and IT infrastructure and executing the deployments in pre-production and production environments. They both report an ever-increasing backlog of release planning and business priorities that currently remain undone. The release window is shrinking and the volume and velocity of deployment requests simply do not fit. Moreover, there is a high dependency on the release analysts and DBAs as they are working around the clock, becoming dangerously close to burning out.
We should reward where it makes sense, but we also need to ensure we optimize our release processes and practices. And I now know of two release analysts who would welcome their weekends back.