• Hi,

    I use default permalinks.

    Cpu usage on average 10-20% , global load 0.30 to 0.40 with sometimes traffic spikes.

    To get a better SEO result, I wanted to enable fancy permalinks structure with Year/month/day/Posttitle.
    After letting it run like that for about 16 hours (second test run, one test run was also done in October 2007, same results).

    Tested with:
    1) normal setup: Wp-Super cache Half On, default permalinks
    2) Fancy permalinks , WP – Super cache FULL ON
    3) Fancy permalinks , WP-Super Cache HALF ON
    4) Fancy permalinks , WP-Super Cache OFF

    Results:
    1) CPU 10-15% ; Load average 0.07-0.50
    2) CPU 60-70% ; Load average 1.20-3.50 (run in this mode for about 10 hours so wp super cache had time to create all cached pages (wp super cache did its job, server load dropped with about 30-40% after creating all pages in a timespan of 4-5 hours)
    3) Same cpu and load average as in 2)
    4) CPU 70-80% ; Load average went to 6.00-9.00

    So I can only conclude that having fancy permalinks enabled costs you up to 4 times more CPU power.

    Remark: WP super cache was tested in version 0.8.5 and 0.8.7 , everything was checked multiple times, cached pages were created and present in the correct folders/subfolders . Check on site itself gave the right Page server by cache notes.
    No traffic spikes occurred, just my average traffic

    So Can this be right? Since I really would like to have fancy permalinks for SEO purposes…. (but having to upgrade my server can easily cost my up to 25-50$ a month extra)

Viewing 15 replies - 1 through 15 (of 24 total)
  • I use pretty permalinks on all of my sites, I think they’re great for humans and search engines!

    I have never noticed a difference in performance using custom permalinks. I would think that you would have to have a huge amount of traffic to notice any extra load, and even then, I wouldn’t expect it to add up to very much.

    I’m afraid I don’t have any good insight for you, but I would suggest recommend broadening your list of suspects. My feeling is that pretty permalinks wouldn’t do this unless your pages have like a thousand links on them 🙂

    Thread Starter loller6661

    (@loller6661)

    Ow ok, well it has to be something.

    Either something is still fishy with my wp-super cache setup or its the fancy permalinks. Since If I put fancy permalinks off my server loads drops back to the normal load like it was over the past 1,5 year ; and I tested it with all sorts of combination to see if it was wp super cache not working.

    I think I have like maybe 20 posts a page and 10 archive links and like 7 subpages; on average 25k uniques a day.

    But I hope its just wp super cache, Altho I tested it like the maximum I can.

    One thing that crossed my mind:

    Wp super cache accesses the .htacces file to put in some rules. Now do this rules have to be adapted to the sort of fancy permalinks structure you use? I was using /%year%/%monthnum%/%day%/%postname%/

    (I do not get any errors anywhere, and the cached files are loaded and created and everything is on its place)

    I don’t think that Super Cache modifies what is already in the .htaccess file, in my .htaccess files, I have two separate sections, one for the pretty permalinks, and one for Super Cache.

    And actually, according to the install instructions, you have to have fancy permalinks enabled to use Super Cache, so they should work well together.

    I’d be curious to know how your test went on a brand new install of WordPress, with fancy permalinks and WP Super Cache as the only modifications.

    Thread Starter loller6661

    (@loller6661)

    Yea I have them as well, both are exactly what it should be (even automatically edited by the plugins/wordpress)

    Anyhow just did a 40 minute test with the only modification being pretty fancy links, compaired to my other setup which I’m running for 1,5 year (of course i always updated wordpress and wp super cache, all other plugins were there from the beginning as well: wordpress backup and a contactform)

    Test :
    Wp super cache Half On + Fancy permalinks on

    <!-- Dynamic Page Served (once) in 0.671 seconds -->
    <!-- Cached page generated by WP-Super-Cache on 2009-01-10 23:53:33 -->
    <!-- Compression = gzip -->
    WP-Cache
    
        * 313 cached pages
        * 0 expired pages

    server load 4.50 for about 5 minutes then dropped to 2.00 ; so about 5-10 times my normal server load…(test was run for 40 minutes)

    Thread Starter loller6661

    (@loller6661)

    Ow right, I also did this test 3 months ago; but changed back to default permalinks because of the server load and lack of time to look into it.

    Just finished reversing back to default:

    Server load 0.20 after 5 minutes the cache was deleted

    <!-- Dynamic Page Served (once) in 0.589 seconds -->
    <!-- Cached page generated by WP-Super-Cache on 2009-01-11 00:46:33 -->
    <!-- Compression = gzip -->

    To illustrate:

    http://img144.imagevenue.com/img.php?image=31353_graph_122_805lo.JPG

    You can clearly see when Fancy permalinks was on

    `

    Wow, that doesn’t leave much to chance if those were the only changes!

    Is there anything non-standard about your web server setup that could be affecting your performance? Are you running a pretty typical LAMP stack?

    Thread Starter loller6661

    (@loller6661)

    Server Version: Apache/2.0.63 (Unix) mod_ssl/2.0.63 OpenSSL/0.9.8b mod_auth_passthrough/2.1 mod_bwlimited/1.4 FrontPage/5.0.2.2635 PHP/5.2.5

    cpu Celeron 2,4Ghz , 1 GB ram , cPanel 11.24.4-R32603 – WHM 11.24.2 – X 3.9

    Server load 0.20 after 5 minutes the cache was deleted

    So you’re saying that it does perform better on a default install?

    Thread Starter loller6661

    (@loller6661)

    Yep, server load is like 5-10 times less when you have default permalinks enabled

    You said that you were using permalinks with this structure: “year/month/day/post-name”.

    Myself, I have always used just “post-name” for permalinks. I don’t see how it would make much difference, but have you tried pretty permalinks with less variables?

    Thread Starter loller6661

    (@loller6661)

    I tried the numeric option for 5 minutes , server load went up the same as well. But I will test it tomorrow for a longer period.

    (I need more variables then only the title since 10 blog posts per day are created, and otherwise I will have a lot of trouble with double names)

    Hmm, and wordpress backup and contactform are the only other plugins you are using?

    Do you having anything out of the ordinary going on in your theme? How many queries does a pageload generate?

    Thread Starter loller6661

    (@loller6661)

    I have http://www.mangoorange.com/resources/i3theme/ theme.

    Object type	Size (bytes)
    HTML:	11826
    HTML Images:	330646
    CSS Images:	100577
    Total Images:	431223
    Javascript:	35956
    CSS:	4646
    Multimedia:	0
    External Object	QTY
    Total HTML:	1
    Total HTML Images:	61
    Total CSS Images:	40
    Total Images:	101
    Total Scripts:	7
    Total CSS imports:	3
    Total Frames:	0
    Total Iframes:	7

    Is that stats for the whole site? I’m wondering how many queries are made to load a single page.

    The default theme has some code in the footer so that you can see how many queries the page took to load:

    <!-- <?php echo get_num_queries(); ?> queries. <?php timer_stop(1); ?> seconds. -->

    Have you tested the load of your site with the default theme?

    Thread Starter loller6661

    (@loller6661)

    No, but I know the load is like 3-4 times higher with this theme (test done like about 1 year ago when I switched themes)

    <!-- 17 queries. 0.718 seconds. -->

    with default theme load went up the first 5 minutes from 0.26 to 1.30.
    After 5 minutes: it started to lower itself to 0.50 (pretty sure it will drop even more when the caching is more done but I stopped the test since default theme kinda screw ups my site)

Viewing 15 replies - 1 through 15 (of 24 total)
  • The topic ‘Fancy Permalinks VS Server Load’ is closed to new replies.