ECS still feels like a preview, but it's coming along! We might blog about our experiences there in the near future. Overall it has a good feel to it, other than some seemingly odd configuration deviations which turned out to be abstractions made for "future flexibility".
I'm using ECS to run a small cluster of 25 odd microservice containers of three different categories spread out over a number of instances.
I'd second your opinion that it still feels like a preview. There are some rough edges that we've encountered such as a mistyped tag name causing ECS to hang forever in the pending state, as it didn't properly catch the error condition of not being able to pull the nonexistent image tag from our Dockerhub account.
One thing that I think ECS really highlights is the need for an ELB service specifically for docker containers. I'd love to be able to run multiple web containers on the same instance and load balance incoming requests to multiple containers on the same machine. Right now because of the limitations of ELB listeners we are instead running one beefy web container per instance with lots of CPU and memory allocation, and one node.js process per core running inside the container. Then ELB load balances requests across our instances, but each instance is still running only one web container.
ECS still brings benefit for deploying and updating web containers, but the true power lies in its management of worker and other background service containers, and the ability to shuffle them around and disperse them wherever there are free resources with very little administration required.
Cloud Foundry does what you want -- push an app, see it run, get load balancing and log aggregation out of the box[0]. And healthchecks, and buildpacks, and so on and so forth.
CF runs on AWS.
You can also play with Lattice, which is a cut down version intended to support fast development cycles[1].
It would be great to hear more about I/O performance on ECS. A while back I ran I/O tests on another Docker-based container service running on Softlayer(Bluemix) and I could get near-native level performance because that environment had Docker containers running on bare-metal hardware. More here: http://www.cloudswithcarl.com/?p=63
