I have heard from many CTOs, CSAs and VP Devs that while they need to be more proactive, its too hard to enforce proactive use of profilers within development teams. Therefore we found that the only thing that works is automation of performance tests in CI environments with continuous feedback to the team and development management.

When asked “where to start” in CI, I found further that its very helpful to take service level goals from production and bring them back to engineering into CI. Thats what I then call performance engineering. If an applicatio is already in production, then another helpful approach is to e.g. use dynaTrace Production Edition to capture business transactions, prioritize them by business criticality and performance weaknesses and take the top 5 into CI for proactive performance management.