I am really quite tired, but it appears I will automate your cloud/VM infrastructure for food. Or money. Preferably money, now that I think about it.
Average-length week, shiny new learnings. Devops Borat is worryingly accurate.
Perhaps I should have saved that title for a more interesting post. Who can say?
Average-length week, shiny new learnings. Devops Borat is worryingly accurate.
Perhaps I should have saved that title for a more interesting post. Who can say?
no subject
Date: 2012-09-21 09:34 pm (UTC)no subject
Date: 2012-09-22 11:13 am (UTC)Common-log format needs throwing away. Webservers should be emitting JSON, which gets hoofed up the wire to an analytics rig with lots of disk. That way, your VMs waste less disk and i/o bandwidth because they're writing/storing logs. And then also running out of disk less often and thus not bothering you at 4am.
You always monitor the wrong things. Or rather, you monitor a set of things and then fix the obvious breakages because you can now spot that deploy 'blah' made a CPU run hot so that's the place to start optimising the code. The next time something properly breaks, it's in a thing that you weren't looking at. So you instrument that bit too, and... Soon you are drinking from a firehose.
no subject
Date: 2012-09-22 11:33 pm (UTC)no subject
Date: 2012-09-23 08:25 am (UTC)[FX: Google]
Ah. Well, there you go.
no subject
Date: 2012-09-22 09:49 am (UTC)