I work in AWS’s Serverless group, and in the process of pulling together my presentation at re:Invent, discovered that I have a lot of opinions on the subject and, while they may well be wrong, are at least well-informed. You can watch that YouTube, but who’s got an hour to spare? And anyhow, blogging’s really my favorite medium, so here we go. If I tried to glom them all together into one mega-essay it’d be brutally long, so let’s go short-form.
The Serverless Fragments · “Serverless Fragment” has five syllables and ongoing doesn’t have that much room at the top of the page for the title, so let’s say “SF”.
SF 1: What is serverless? Tl;dr: It’s when you don’t have to reserve capacity before you start. But it’s not a binary condition.
SF 2: Why serverless? Tl;dr: Frugality, Security, Elasticity. That part’s a no-brainer, but will you also get better software?
SF 3: Serverless everything? Tl;dr: Nope, serverless where possible. To start with, think about control planes and data planes.
SF 4: Serverless latency? Tl;dr: It’s complicated. And, there are things to do about it.
SF 5: Serverless bills? Tl;dr: Not always lower. But it depends how you count it.