Although AWS recently published their experimental public roadmap for AWS Container Services.
The company can still surprise us during a global pandemic by suddenly announcing the launch of the new Fargate run profile version 1.4. Which brings us some interesting new features and also some big changes under the hood.
For those who do not know, AWS Fargate is a serverless compute engine for containers. It allows you to run, auto-scale and self-heal your containers without the hassle of managing a generic infrastructure. It is limited in multiple ways to it's bigger brother k8s, but the learning curve is low and I personally believe for most use-cases good enough.
It seems version 1.4 is immediately available *or at least in all the EU regions we operate). So what is the big news ?
Persistent volume support
Well first one of the big drawbacks about ECS Fargate was the lack of support for persistent volumes. Personally I am not a big fan of persistency in containers, but that's for another discussion. Anyway this lack there of, is now ended. You can now mount Elastic File System (EFS) volumes direclty in ECS Fargate tasks and since it's based upon NFS (Network File System) this is a widely adopted, supported and a very flexible solution.
Adding extra network monitoring
By adding a new Fargate agent in the profile enabling access to more network performance metrics, we are now able to see more details of network performance in Container Insight. When deploying a task from 1.3 to 1.4 you can quickly notice the new network transmit (TX) and receive (RX) data. Which is a feature long available for EC2 instances but now also for additional monitoring of our high traffic containers and tasks.
No more docker
As explained in the post here in more technical detail the new profile 1.4 will also replace the Docker Engine with containerd as execution engine. Containerd is a 100% open-source industry-standard container runtime backed by the cloud native computing foundation. Our applications will see no differences in how they are executed. But containerd helps AWS to improve their data plane architecture by simplifying it and making it less dependable of a commercial company like Docker.
Although this will have no impact for us developers it is actually a big thing. It shows one again how the industry is moving away from Docker.