Latency testing

Locate delay in one leg of a regional request.

Compare connection setup and transfer

A first-party status page returns a small fixed payload. Measure a new connection and a reused connection separately from the same regional test client. Record sample size, failures, response size and the exact timing boundary used in each series.

How to run the review

Compare small fixed payloads and group cold and warm connections. Retain unsuccessful attempts beside median and tail timings.

Network selection

Compare the characteristics of datacenter proxies with the test's geographic and session requirements. Check the planned region in the coverage directory.

Session setup

Use one datacenter exit per measurement series. Compare cold and warm requests separately, and label a new series whenever the exit, payload or destination changes.

Frequently asked questions

Should a timed-out request receive an invented latency value?

Report the timeout separately and retain the configured deadline. Calculate successful-request timings from observed durations and show the completion rate beside them.

Can we compare two markets tested at different hours?

Document the time difference and any known origin load. Matched observation windows and consistent cache state make a regional comparison easier to interpret.