2 ms·
Another difficulty in DevOps is that you're never really off. If deployments can only happen in the wee hours of the morning or you have to answer that call abo
by JustSomeNobody 10y ago
Another difficulty in DevOps is that you're never really off. If deployments can only happen in the wee hours of the morning or you have to answer that call about an issue with the system and you're spread over several time zones, you are always "on".
- memracom 10y agoThis is a warped idea of what is actually happening. It is not about making developers take on old-style system operations tasks. It is about eliminating those old-style tasks. Nowadays, this is really possible. We can build self-healing software and even something old like Nagios can alleviate most of the on-call type work of the past.
- jankedeen 10y agoThere is no 'self-healing' system. Nothing can detect a 'real' failure as opposed to a counterfeit failure without heuristics that depend on an interpretation that is inherently fallible. Ops was about discipline and method. Restrict the problem set to known values. The one thing that I see these days is that developers don't have a fucking clue about ops but they do see that the basic task responsibility is toilsome, counter-creative and that 90% of it can be automated by reliance on coded procedures in write once argot that hopefully is editable and doesn't give up the family jewels when posted in pub github.