Page MenuHomePhabricator

decide on an aggregation function to combine multiple probes into a single measurement
Open, Needs TriagePublic

Description

For both the NEL latency data and the Probenet latency data, from any given user, we'll have reports for a set of N pulses towards each datacenter.

We expect the first pulse to report an artificially high latency number (as it includes all the overhead of TCP + TLS handshaking).

We also expect there to be many other sources of noise in any/all of the pulses.

So far we've thought about taking the minimum latency reported, or the 10%ile, or doing a trimmed mean. After we have a non-trivial amount of data, we should settle on one 'best' strategy.

Event Timeline

Hi @CDanis, can you please associate one or more active project tags with this task (via the Add Action...Change Project Tags dropdown)? That will allow to see a task when looking at project workboards or searching for tasks in certain projects, and get notified about a task when watching a related project tag. Thanks!

Since in every probe, we are making 3 requests to each data center and the first one has artificially high latency. That means around 33 % of the probes will have to be treated differently. So I suggest we try 33 %ile, 50 %ile and 66 %ile also.