- Newest
- Most votes
- Most comments
One solution is to preload the container image in the AMI itself, which can significantly reduce the startup delay . When the image is present in the instance, ECS agent only needs to compare checksums with the ECS repository, starts in less than a second for image download . For tasks running frequent jobs, you can implement a persistent container that runs continuously rather than creating new containers each time. This can be achieved by using a task scheduler like node-cron within the main application code, allowing the same container instance to be reused for multiple job executions
answered 2 years ago
Hi,
You may be interested in "Reducing AWS Fargate Startup Times with zstd Compressed Container Images": see https://aws.amazon.com/blogs/containers/reducing-aws-fargate-startup-times-with-zstd-compressed-container-images/
Also the re:Invent presentation "Reducing AWS Fargate startup times by lazy loading container images" proposes different paths to explore to reduce startup time: https://d1.awsstatic.com/events/Summits/reinvent2023/CON307_Reducing-AWS-Fargate-startup-times-by-lazy-loading-container-images.pdf
Best,
Didier
Relevant content
asked 6 years ago
asked 2 years ago
- AWS OFFICIALUpdated a year ago
