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
Two reasons they're slower, in my experience:
awsvpc network mode attaches an ENI to the task. When it has to do this to a Lambda, if the Lambda is running in a VPC, it is known to dramatically increase the initial spin up time.
Docker image size also affects startup time, since the image will usually need to be downloaded to whatever hidden host for a task to launch. I've done some benchmarking with a small 200MB container and a 2.5GB container. The former did start up quicker.
You can't do much about awsvpc, since Fargate requires it. Shrinking down that image would be your next biggest impact.