Hi We have a flink streaming app which reads messages from kafka. These messages goes to an asyncio function which makes REST calls to different endpoints. There can be 8-10 types of messages and each type is a different endpoint. The problem I am trying to solve is to somehow isolate any issues in an endpoint. For example if one endpoint is very slow it impacts calls for all other messages. Any ideas on making the app resilient so that it is not severely impacted in these cases?