HTTP1 downstream connections are either waiting for the first byte of the request headers, parsing request headers or handling the active request. We benefit from relatively long timeout for connections with no active request as it allows for effective prefetching and connection reuse. We do expect a relatively short time between when the first and last byte of the request headers arrive which could be covered by a shorter timeout.

Similar logic could also be applied to H2 connections with 0 active requests.

The timeouts for idle connections with no active requests and connections parsing headers could be reduced under could be reduced based on memory pressure to increase resiliency to memory exhaustion attacks and improving the proxy's ability to accept legitimate connections/requests. Config strawman: min/max timeouts and high/low memory thresholds at which the timeouts apply or high/low connection count at which timeouts apply.