Envoy has a latency-based adaptive concurrency feature: https://www.envoyproxy.io/docs/envoy/latest/configuration/ht...

Netflix has a blog post for their implementation: https://netflixtechblog.medium.com/performance-under-load-3e...