I've been asked by a number of people "How much does page cache help?"  While I have been able to answer that it does speed things up, it has always bugged me on why some users get a GREAT performance gain, and others only a slight gain.  So today, I ran some queries across three live sites that I have PageCache running on (On The Web, BagsOfCrap, and NBVFC).  Now, you need to note that all three of these sites are running the same server, and are running the same version of Joomla.  Each is running a pretty simular set of components (Joomlaboard, OpenSEF, Community Builder), but has different main components (NBVFC's main part is a gallery2 Image gallery, On The Web and BagsOfCrap are mainly content driven).  The results are rather intersting...

        I want to start off by describing my traffic patterns on each site.  Ircmaxell.com and Nbvfc.org are both steady.  They are not very busy, but they do see traffic on a regular basis.  BagsOfCrap, on the otherhand is EXTREEMLY bursty.  28 days out of the month it might see 500 hits per day.  But on those other 2 days, the server will max out due to the traffic.  From May 30 to June 1st, my server pushed out over 50,000,000 requests (Not bad for a virtual server with 750 megs garunteed ram!!!).  Now, with that in mind, take a look at this graph. 

        This graph show the hit distobution of each site.  Blue is the number of cached hits, Red is the number of Disabled hits, and Yellow is the number of uncached hits.  As you can see, BagsOfCrap looks like it bennifits much more from PageCache than either of the others.  Is that true?  Lets find out!  This next graph shows the average time spent generating a page (for each type of hit).

        As you can see, there is not much difference between Ircmaxell.com and BagsOfCrap.com, but Nbvfc is WAY off.  I think this has partly to do with the image gallery, but I cannot be sure without digging into the code.   This shows a clear trend.  On all sites, pages loaded at least twice as fast, and in one case 25 times faster when cached.  One thing to note, is that caching the pages does take a little time (as you can see from the difference between disabled times and uncached times). 

        There is one thing we can difinitively say about these results.  You need to try it for yourself!  On a site like Nbvfc.org, it makes a HUGE difference in load times.  On a site like this, it does make a difference, but not that big.  But, there is also the resources usage to concider before saying that it's not worth it.  On BagsOfCrap.com, I could not support the number of hits without the caching!

NOTE:  If you would like to know the query I used to generate the data, here it is!

SELECT SUM(ctime) AS ctime, SUM(chit) AS chit, SUM(ctime)/SUM(chit) AS cavg, SUM(dtime) AS dtime, SUM(dhit) AS dhit, SUM(dtime) / SUM(dhit) AS davg,  SUM(utime) AS utime, SUM(uhit) AS uhit, SUM(utime)/SUM(uhit) AS uavg GROUP BY null

TrackBack URI for this entry

Comments (1)

Great component!
written by Sean, June 12, 2007

I've noticed a huge boost in performance after installing Page Cache. My site was being shut down by my hosting provider because of resource usage before installing the component. Now the pages load way faster (except in the forum which is excluded) and the server load is reduced. Thanks a bunch!

Name

Email

Website

Title

Comment

Write the displayed characters

Add Comment