So you’ve deployed your app, launched your site, and gone live in prod: so far, so good. But what happens when people actually use it, when you find out where your IaaS rate-limits you, and when you can’t just throw (virtual) hardware at your software problems?
Let’s explore the practical realities in that vast, uncharted space between “waiting for the change control board is probably fine” and “we just implemented a novel time series data store in our spare time.”
If you’re coming to this talk, you may be well past the former, while the latter remains strictly aspirational. What matters these days in operations when you’re dockering all the Dockers, going serverless with lambda functions, or scaling beyond what you’ve seen before? Getting the abstractions right is a non-trivial task.
I’ll chat about the new stuff worth running in production, dispel myths (NoOps is not a thing; serverless still has servers), and answer questions (spoiler alert: the answer is “it depends”). If you’ve put in a non-zero number of hours on the pager and still can joke about it, you know that a Someone Else’s Problem field lasts just long enough for it to become our problem.
Let’s talk solutions.