Is anyone here using HashiCorp products at some significant scale? I like the general idea they are using (because I have had the same idea) of building nice, reusable, components and belting out products based on them. At the same time, I know of nobody actually using them so I’m not sure how well they actually work.
For this particular product, scheduling at scale is something whole companies are based around this and it’s, more or less, the core competence at Google. I’m not sure how a company like HashiCorp can compete, at scale, with that.
We deployed Consul at scale at SoundCloud. In general it performed as advertised. We had to stick several DNS recursors (effectively caching layers) in front of the agents to meet our DNS load requirements.
We use packer heavily to build AMIs on AWS.
We’ve started using Terraform and are very happy with it, despite its quite steep learning curve when operating multiple environments.