The production ROR API currently uses a single AWS Elasticsearch instance, 2 containerized application instances and a single application load balancer instance. During periods of heavy use (ie, every day at midnight UTC), CPU in the OpenSearch instance becomes a bottleneck and requests are rejected.
During 2023, we plan to add the ability to run multiple Elasticsearch instances simultaneously and to scale automatically services automatically, based on demand.
The production ROR API currently uses a single AWS Elasticsearch instance, 2 containerized application instances and a single application load balancer instance. During periods of heavy use (ie, every day at midnight UTC), CPU in the OpenSearch instance becomes a bottleneck and requests are rejected.
During 2023, we plan to add the ability to run multiple Elasticsearch instances simultaneously and to scale automatically services automatically, based on demand.