Rendered with Parsoid
From Wikitech
Latest comment: 7 years ago by HaeB in topic "Latency is bimodal", and caching

"Latency is bimodal", and caching

... vs. actual

There was some related discussion at phab:T125414. In short, while the remarks about caching seeem valid in principle, in the actual data from enwiki it is not true that "the distribution has two peaks". While it may still be bimodal in the modified sense that it the sum of two simpler distributions (or actually four of them since there are two different kinds of caching, client-side in the browser cache and server-side on the Varnishes, a fact that's missing from this page too), that effect seems weak enough here to regard the median as robust enough metric, contrary to the recommendations here. Regards, HaeB (talk) 22:26, 11 August 2016 (UTC)Reply

Given the percentage of our backend caching (i.e. we end up serving mostly from varnish >90% of the time) the signal you are most likely to see is the bimodality due to frontend caching. So, no, you would not see four modes easily in our data. In enwiki you are bundling up all users all countries and thus the bimodality is affected by connection speeds. Makes sense? That doesn't mean that the median is a good representation overall of our user base. Reduce your analysis to users with similar connections and locations and you will see the difference. These are well known issues that affect any website deployed globally, not just enwiki.

Some literature: