The story https://issues.redhat.com/browse/NE-991 produced initial results for haproxy vs. envoy, but there are massive delta's between two comparable runs. We are planning to have Perf & Scale produce standardized and peer reviewed results in https://issues.redhat.com/browse/NE-1089.

After NE-1089 is complete, this story is to understand and recommend changes to haproxy vs. envoy testing techniques and parameters as well as pertinent proxy configuration. We also want to understand generally why there are deltas between haproxy and envoy.

After understanding the delta, produce documentation for customers on what Envoy is useful for (e.g. not useful for high performance applications).