AWS Summit 2019: Microservices on AWS: Architectural Patterns and Best Practices by Sascha Moellering, AWS


Microservices in a deep look

Options for architecting your microservices:

  • agility can increase if the impact of the release is small;
  • small identifiable pieces, which are doing one thing well; only the contact is important, not a language of choices for each of microservice;
  • fault-tolerance delivery.

Containers: Amazon ECS, EKS, AWS Fargate.

Serverless AWS Lambda

What if I can not decide?

Do you really need microservices? Why not to go serverless.. ​

This example shows how importing the code happening. Lambda will keep it in the container. This code executes on the cold start. ​

Containers best practices

  1. Reduce size of your docker images. Time to pull images from docker registry impacts your application startup time.
  2. Optimize pods: keep them lightweight, admission controllers add overhead!
  3. Put a limit on the max resources of a pod.

Conclusions

There is no silver bullet! ​