As a manger/lead engineer/program manager how did you go about implementing Devops culture at your workplace? Specifically trying to identify pieces related to people and process . As an Engineer one can follow good design/architecture , use version control,pull request , CI/CD,release management , testing and following engineering standards. What are the other organizational pieces that are involved?
1. Put a blameless post-mortem culture in place. This means celebrating when people are transparent on how things messed up and never ever ever shooting the messenger, but patting them in the back instead. 2. Take post-mortems seriously and come up with real action items that are system level improvements. E.g. “do more thorough code reviews” —> not a real action item, as it does not have system level improvements. “Automatically block landing code without test coverage or deploying with failing tests” —> good action item, with systemic improvements. 3. Prioritise post-mortem reviews, with experienced engineers & respected managers being present and wanting to understand & challenging the root cause & action items 4. Prioritise executing the action items ahead of other work. Treat it with similar priority like an ongoing outage. You do the above for 6 months and you will naturally have things like alerting, monitoring, CI, CD and other engineering best practices in place. Also, if you don’t have this culture, you need to walk the walk over talk the talk. So roll up your sleeves and use your influence on your team to start doing the above and you investing your time & focus heavily, prioritising it over a lot of the other stuff that your company values right now (like shipping immediate value to the business). Good luck!
You ever want an SRE referral you send me a PM.
What is Devops culture? As a manager, I look for ways to maximize productivity.
Hire some ops that can dev
Checkout The Phoenix Project. It’s a quick read and teaches the fundamentals for setting up a devops culture. It also tells you what not to do with various parables. I found it incredibly helpful when in a similar situation.
Thanks ! Last 40 pages is a gem
Have metrics for the things you care about. Have alerts that wake people up for things that if they break would affect your customers. Push to make every page important. Push to fix every page by fixing the root cause. By making that service self healing. You should get woken up for anything that could affect your customers, but you shouldn’t get woken up by the same thing twice - if you fixed it right the first time.
I hate companies where too many people try to modify the “process” like that’s what senior people do. Just do your damn job