From investigating various possibilities (https://jira.camunda.com/browse/OPT-3153), it seems like we can achieve what we want by using a QoS filter: https://www.eclipse.org/jetty/documentation/current/qos-filter.html. We should add this to apply to the ingestion API which will limit the number of ingestion requests the server handles at a given time

ATs:

• A quality of service filter limits the number of requests to the ingestion API
• The number of requests that can be handled is configurable
• The performance of the rest of the API isn't impacted by high loads to the ingestion API
• Configuration documentation is updated