We can measure a tight upper bound on router+network latency.
We can also measure in-dyno queueing latency (between webxd
and the app). We can also provide flow control statistics (e.g.
how many backend retries did the router perform; how long did
the request sit in the router queue before being tried; etc).
We can measure a tight upper bound on router+network latency. We can also measure in-dyno queueing latency (between webxd and the app). We can also provide flow control statistics (e.g. how many backend retries did the router perform; how long did the request sit in the router queue before being tried; etc).