Aug 28, 2017 · 1 min read
For what it’s worth, We use KOPS to spin up a kubernetes cluster in AWS with it’s own VPC pretty quickly and painlessly. We then deploy a portion of our AWS lambdas to that account. On top of that, we’ve got automation in place to create vpc peering, etc. This avoids rate limits and provides a smaller blast radius when things go wrong, but increases the overall AWS spend in ways I don’t feel we have any great rules of thumb for. We are also exploring using kubeless for the most problematic AWS lambdas. I did very much appreciate your article, by the way, and I’m interested in following your journey.
