As a service provider we should not be reliant on the flowing traffic to measure response latency of cid.contact. This is because if there is no traffic flowing we will be blind to increase in response latency, say, as a result of code changes.
Run a service, ideally in different region which hits cid.contact on random known CIDs and report metrics on response latency.
As a service provider we should not be reliant on the flowing traffic to measure response latency of cid.contact. This is because if there is no traffic flowing we will be blind to increase in response latency, say, as a result of code changes.
Run a service, ideally in different region which hits cid.contact on random known CIDs and report metrics on response latency.