2 posts tagged with "mttr"

Two production outages, same root cause: a bad config push that crashed a payments service. Team A spent 2 hours 14 minutes restoring service. Team B was back in 6 minutes. Team B's MTTR wasn't lower because they had smarter engineers. They had a one-command rollback rehearsed monthly, a runbook pinned in the on-call channel, and write access to production already granted to the responder. That 134-minute gap is what MTTR measures, and what separates the DORA 2023 State of DevOps Report elite cluster from everyone else.

Google's Site Reliability Engineering book (2016) popularized a counterintuitive principle: accept failure as inevitable and invest in recovery speed. The DORA research confirmed it with data — the difference between elite and low-performing teams isn't that elite teams have fewer incidents. It's that they recover in under an hour instead of under a week. Every engineering organization invests in preventing failures. Fewer invest in recovering from them quickly. The data says this is backwards.