#DevOpsDays#DevOpsDaysMSP@alicegoldfuss: 4) Containers need headcount. Containers get shoved onto ops, because it seems like what they’re doing, but it’s not like “a new Jenkins”. It’s building out a datacenter in city no one has ever been to, with hardware they’ve never seen.
#DevOpsDays#DevOpsDaysMSP@greenreedtech: We want to use machine learning to give us more and faster data about how our production is behaving, not to replicate what I can do myself?
We don’t want our deployments to be like playing frogger with blinders on; automation can help us fill in the missing context. @GreenReedTech#devopsdays
Distributed systems failures make the best stories. I love how @johnrengelman is slowing revealing the complexity of the failure mode. Even describing non technical contributing factors like Engineers trying to restore service to their services by redeploying them #devopsdays
In order to troubleshoot in a large, complex distributed system, you have to have levers... You have to be able to isolate functions. @johnrengelman#devopsdays
#DevOpsDays#DevOpsDaysMSP@lanooba: 2) Be mindful about data - Start with an explicit data collection strategy. Conduct exploratory data analysis. Incorporate open source data sets into your models. Use crowdsourcing to fill gaps. (hire a diverse team to spot things you miss)