Re: [Multimedia] Pre-rendering thumbnails on the back-end

Gergo Tisza, 09/07/2014 00:55:

There is now: https://bugzilla.wikimedia.org/show_bug.cgi?id=67698

Thank you! Following up on the "Swift capacity" thread, a doubt developed in the last few days in my brain:

Gergo Tisza, 06/07/2014 23:08:

    Media Viewer could benefit greatly from this performance-wise.
    As seen on this graph, the launch to all wikis affected the
    average considerably, since users started hitting a lot of
    images that didn't have Media Viewer-sized thumbnails yet:

http://multimedia-metrics.wmflabs.org/dashboards/mmv#overall_network_perform...

This looks pretty bad. Thanks for calling it out.

I don't think it's especially bad; there is a spike after the rollout (it can be seen more clearly if you scroll down to the imagemiss stats) which lasts about five says, other than that it's just probably the effect of rolling out to new userbases which have on average much worse network conditions then the Europe/USA based ones.

If you look at wikis to which we have rolled out earlier, e.g.

http://multimedia-metrics.wmflabs.org/dashboards/mmv_enwiki#overall_network_...

there is no change at all.

Which is not to say the lack of pregenerated thumbnails is not a serious problem (I just don't think it got any worse recently). Comparing the global imagehit and imagemiss stats, the lack of pregeneration affects about 20% of the requests, and costs about 730ms (an extra 85% loading time) for the median user.

This does make me wonder, what do those graphs *actually* measure? Are they really measuring a comparable/representative set of image requests from which we can infer that one or the other method is intrinsically faster than the other, or are they measuring different things e.g. for some selection bias?

Nemo