Hi! Thanks for your feedback on the ecosystem map!
AWS ECS: AWS ECS is not on the "Containers and Orchestration" category line because the inclusion criteria is for technology that you can download and run anywhere. Happy to discuss further whether this should be changed.
Haschicorp: This was an oversight, it will be added in the next rev.
gRPC: Protocols such as gRPC, JSON-RPC, etc. are not currently part of the inclusion criteria for the Specifications line.
ECS is awful!! Their registry (ECR) is pretty good and painless to set up. But their container service is just beyond atrocious, and not in a "this is a young product" kind of way. I was forced to build on top of ECS earlier this year and absolutely hated it.
It not being my project, merely a project that I built, I didn't have the choice to use something else, so I slogged through ECS. So basically, just off the top of my head... AWS sucked all the fun out of Docker. The ECS agent on your ECS nodes has (or had, it might be fixed, but who knows) the nasty habit of randomly becoming unresponsive, and so the ECS agent gets restarted in order to re-establish contact with AWS, which also means that it restarts the Docker daemon. That means all of your containers get restarted. Amazon also funnels you into using a single container type per EC2 instance. It's not impossible to use a single EC2 instance for multiple containers, but if you desire to run multiple instances of one specific kind of container on one node, ECS doesn't make the implementation easy for you at all. ECS also defines entirely new terminology. It's pretty obvious that AWS wanted to add a layer of abstraction so that they can swap out Docker with some other container technology, which is fine, but you're going to be slogging through Amazon's infamously-bad documentation trying to get a handle on their particular wording. Their documentation for ECS is just awful, like it is for almost everything else they make. I am a regular in IRC, and ##aws on Freenode is almost always totally useless. Those are some things I can offer up right now, and I definitely ran into other problems while working with ECS that I can't think of at the moment. It was one blocking problem after another, and I've been doing systems and operations and software engineering for twenty years, and very familiar with Docker's containerization and plenty of other orchestration tools. It was way worse than it should have been.
That's the mistake I found after looking at the map for 10 seconds. Many companies are running production grade services on ECS and its exclusion should be a mistake than intentional.
My question is, how did Rancher Labs get on the list, but Hashicorp didn't?
Also, Google deserves a link to specifications with gRPC, IMHO.