Customize

Hello SPDY

Discussion in 'Site News and Announcements' started by sue, Jul 27, 2012.

  1. sue Administrator

    Whyweprotest is now part of a wider cloudflare beta test and benefits from SPDY support. What is SPDY? you might ask, SPDY is an experimental networking protocol developed primarily at Google for transporting web content.

    The goal of SPDY is to reduce web page load time. Which in all honesty might be hard to notice considering the recent improvements in site responsiveness.

    SPDY is supported by the browsers Google Chrome/Chromium and Firefox (version 11+).

    While unlikely: If you encounter any out of the ordinary/persistent pageload issues or bugs, please let me know. Either in this thread or by email: sue@whyweprotest.net

    Once SPDY reaches stable on nginx It will be deployed on the origin servers, hopefully improving page load times even further.

    Screenshot from 2012-07-27 15:29:27.png
    • Like Like x 2
    • Winner Winner x 2
  2. Incredulicide Member

    Awesome stuff! :)
    • Agree Agree x 1
  3. grebe Member

    • Like Like x 1
    • Funny Funny x 1
  4. Kilia Member

    Thank you for the notification, sue. Much appreciated with what you are doing to make this a better place.
  5. sue Administrator

    Well in this case it was a matter of answering a few emails and pushing a button :)

    As for SPDY on the origin side, word is that it wouldn't (in it's current state) improve performance much more, so will hold off on implementing that.
  6. I have been having lots of partial page loads since you did this, like the last time you tried SPDY?
  7. sue Administrator

    The partial page load issues dating from a few weeks ago had nothing to do with SPDY. Could you point out on what pages you experience those issues?
  8. All of them.

    I remember you saying you did something several months ago to make the pages load faster and I had the same problem.
  9. sue Administrator

    that was lazy loading, nothing changed in recent days, how long has this been happening?
  10. Since Friday.
  11. amaX Member

    SPDY makes me miss housespider. anyway, well done on the latest computer thingie that will do whatever it does better than ever before. thanks for taking the time to do the backbreaking work of answering some emails and pushing a button! you must be exhausted. ;)
    • Funny Funny x 1
  12. sue Administrator

    I'll look into this, tonight, tomorrow and Wednesday morning I won't have much time for it but i'll try to narrow it down / fix it by end of this week.
    • Like Like x 1
  13. Ann O'Nymous Member

  14. sue Administrator

    Bumped/tweaked caching up a bit, removed some unused addons. It might resolve some of your performance issues, No dice on proper - large scale performance tho.

    Explanation:

    I've been fighting with caching for some time now. Here is what it is supposed to look like: (same servers)

    Some results: cultmap.net (optimization works) 1-250 connection rush 1 minute long: load stayed under 0.4, no timeouts, it's as if nothing ever hit.

    Test was run 2 minutes ago and generated 6,750 successful hits in 1.0 min and we transferred 83.48 MB of data in and out of your app. The average hit rate of 107/second translates to about 9,297,239 hits/day.
    http://blitz.io/report/888068a1379c2414ee29a27d3ffb5d15

    Some results: cultmap.net (optimization works) 1-1000 connection rush 1 minute long: load stayed under 1.3 (out of 6 cores, good) Some timeouts ( acceptable )

    23,285 successful hits in 1.0 min and we transferred 291.08 MB of data in and out of your app. The average hit rate of 371/second translates to about 32,091,451 hits/day.

    Some results: cultmap.net (optimization works) 1-5000 connection rush 1 minute long: load stayed under 3.05 (out of 6 cores, good) Some timeouts ( meh, eventho at this point loadbalancing with other servers would definitely be implemented )

    80,286 successful hits in 1.0 min and we transferred 247.86 MB of data in and out of your app. The average hit rate of 1282/second translates to about 110,800,497 hits/day.

    http://blitz.io/report/4fa874ee671bb31df2ba1ec7033af4f5


    So basically what this means, when caching works properly: there's little sites that could generate enough traffic unnoticed for it to truly impact services.

    However, the main wwp site is setup slightly differently from cultmap.net amongst other things: HTTPS/multisite/xenforo/mediawiki

    Meaning, there are multilple layers of complexity, it's going to take some time for me to polish this.

    Either way, yes the site was a little unresponsive while I was running those tests. My bad.
    • Like Like x 3

Share This Page

Customize Theme Colors

Close

Choose a color via Color picker or click the predefined style names!

Primary Color :

Secondary Color :
Predefined Skins