Realtime Hosted Service Latency Stats

Note: the live latency benchmarks that used to run on this page are no longer collected — the reporting service that powered them has been retired. The methodology and code remain available in the realtime benchmarks repository.

How the benchmarks worked

The latency was collected and calculated using the code in the realtime benchmarks repository. The benchmarks were executed on every client which navigated to any page on www.leggetter.co.uk. Latency was calculated based on the round-trip time from triggering an event from the client and then receiving that message back — so the latency measured was the latency from the visitor’s web browser at that moment.

Stats were collected for a number of realtime hosted services from Sun Oct 28 2012 onwards.

Why does latency matter?

Quite how important latency is really depends on the use case of your application. If you’re building a trading application then milliseconds really do matter. However the majority of the use cases for web, mobile and Internet of Things applications actually means that getting data within a second is acceptable.

If you want to measure latency for services today, take a look at the realtime benchmarks repository and run the tests in your own controlled environment.