Software engineering · Operations & workflows · Product & design

Latency.

The time taken for an operation or request to produce its response.

Useful when “make it faster” needs a measurable target.

“Measure search latency from the user’s request to the displayed result, including slow requests.”

A closer look

Specify where measurement begins and ends. Server processing time may omit network delay or browser rendering. Averages can hide slow experiences, so a percentile such as the 95th can help describe the slower tail of the distribution.

Put it to work

“Compare median and 95th-percentile latency before and after the change.”

A useful distinction

Latency is time per operation. Throughput is how much work completes per unit of time; improving one does not always improve the other.

Further reading

Monitoring Distributed Systems · Google SRE
Revision 1 · EN