How to speed up deployments on AWS Fargate?

Question

After migrating from EC2 cluster instances to AWS Fargate, I realized that deployments take a lot longer. Before they would take 1-2 minutes, now some deplyoments take up to 5 m

Answer 1

Here's the breakdown of tasks and possible improvements that I've found while researching options to improve my deployment times with ECS Fargate:

Fargate Deployment Overview

Here's a breakdown of what's going on behind the scenes that attribute to the deployment duration:

• Provision the Fargate worker instance
• Provision/attach the ENI
• Download the Docker image
  • Here you have opportunities for improvement:
    • reduce the size of your Docker image
    • Networking throughput is based on the CPU allocations to the Fargate Task - if you allocate more CPU then you get more networking and the image will download faster
• Application Startup time
  • Becomes a factor if your application requires a health check grace period, again effected by CPU allocation

If your task is associated with a load balancer the deployment will also need to pass health checks, and you'll need to account for:

• Load balancer deregistration delay
• Pass health checks: (Health Check Interval * Threshold)

How to deploy Fargate Task updates faster

• Over allocate the CPU
• Reduce the deregistration delay
• Set the health check threshold to 2 and interval to 5 seconds
  • don't forget to account for a health check grace period if your app needs it

My Results

During my testing, I was able to deploy my application that typically takes about 8 minutes w/1024 CPU (1vCPU) in under 4 minutes w/4096 CPU (4vCPU)

Disclaimer

Likely your tasks typically require considerably less CPU and you don't want to be always paying for over-allocating the CPU. So, run your deployment with overallocated resources and then run another deployment right after with the original CPU allocation.

Probably not a solution you want to use for every deployment, but could be a solution for hotfix deployments.