AWS re:Invent 2020: Scaling containers on AWS

Running containers on AWS, especially at scale with 10,000+ containers, can be challenging. Scale comes with many more responsibilities that are often not visible until much later in the process. This session covers different scaling behaviors for several AWS orchestrators (Amazon ECS, Amazon EKS, AWS Fargate, and AWS Lambda) and the patterns that emerge. You hear considerations for both developer velocity and the operational load involved. Cost management is a top priority, and opportunities for cost optimizations and patterns are discussed in depth.

Subscribe:

#AWS #AWSEvents

[javiermoreno]

Great work, love the way you make pauses to create tension and put us in a participative mood. Thanks.

[ranibow]

Great presentation, I don't believe the comparison of ECS and EKS on EC2 is comparable though; adding new nodes is probably the most lengthy part of the autoscaling process. You can't compare the two if you've already pre-sized the autoscaling group for ECS. Awesome presentation otherwise, thanks!

[quincymitchell]

Ha, I thought the green line would have been Lambda