Welcome, Guest. Please Login
 
  HomeHelpSearchLogin FAQ Radified Ghost.Classic Ghost.New Bootable CD Blog  
 
Page Index Toggle Pages: 1
Send Topic Print
HTML compression and Apache mod_gzip (Read 17134 times)
Rad
Radministrator
*****
Offline


Sufferin' succotash

Posts: 4090
Newport Beach, California


Back to top
HTML compression and Apache mod_gzip
Sep 28th, 2008 at 5:44pm
 
Been reading up on web site optimization (WSO): http://www.websiteoptimization.com/

.. which stresses (among other things) small, fast-loading pages, light on gfx.

I ran some tests here: http://www.websiteoptimization.com/services/analyze/

.. which gives a nice breakdown of all web page components » html, css, gfx, javascript, etc.

In the process, I read some things that suggest it's possible to get ~50% compressio using HTML compression .. with things such as mod_gzip, which is an Apache mod:

http://www.bluestream.org/Networking/gzip.htm

http://sourceforge.net/projects/mod-gzip/

I think the Rad VPS uses Apache 2.2.9, which is the most recent version (June 2008), so we should support whatever compression tools are out there.

Anybody know about mod_gzip, or the like?

Apache seems kinda scary to me .. a back art.

I alwasy try to keep most of my pages small, but 50% compression definitely seems worth it. Anything that would make web pages load faster and save bandwidth is a good thing?

Thoughts? Experience? Caveats?

I know web browsers must support compression. If they DON'T, I understand a regular non-compressed page will be sent.
 
WWW  
IP Logged
 

MrMagoo
Übermensch
*****
Offline


Resident Linux Guru

Posts: 1026
Phoenix, AZ (USA)


Back to top
Re: HTML compression and Apache mod_gzip
Reply #1 - Sep 28th, 2008 at 7:24pm
 
Compression saves on bandwidth.  It doesn't always mean that the page loads faster.  It takes some time for the server to compress the file, and some time for the client to decompress it.  If your bandwidth is really low (dialup) then compression would save time because the computer is faster than the connection.  With higher bandwidth like cable and DSL, it could actually be slower to use compression.

I can't tell you how much of a benefit/trade-off there is for html compression.  My experience with real-time compression is mostly with VoIP and SCP file transfers. 

I do know that we recently enabled html compression at work and we are seeing an increase in HTTP 500 errors (which is an "Internal server error"  ..very specific.)

Apache isn't really black magic.  It is fairly straight-forward.  It probably seems complicated because it has lots of options and can be extended by a large number of modules.  Fortunately, the configuration is also modular and stored in a *nearly* human readable text file.  The configuration files often even come with self-documenting comments.  So, it looks big and complicated at first, but it starts to make sense once you dig in and play with it a little.
 
WWW  
IP Logged
 
MrMagoo
Übermensch
*****
Offline


Resident Linux Guru

Posts: 1026
Phoenix, AZ (USA)


Back to top
Re: HTML compression and Apache mod_gzip
Reply #2 - Sep 28th, 2008 at 7:36pm
 
A quick google seems to indicate compression was all the rage back when most users were on dial-up, but I wasn't able to find any recent data on the pros/cons of compression.  I suspect that with home bandwidth speeds growing, compression is less of a boost than it used to be.

I did find one article as late as 2006 where the author ran some tests and saw about a 15% improvement when compression was turned on.

http://newestindustry.org/2006/10/03/performance-improvement-from-caching-and-co...

Compression works better in general for text, so you will see more bandwidth savings and more speed improvements for pages that have more text than pictures.  Since CSS and Java scripts are sent as text, you will benefit from compression there, as well.

That is not to say that I think pages should be all text.  Part of the fun of 2008 vs 1998 is that home internet speeds have increased to the point where we can afford to put a few pictures on our web sites.
 
WWW  
IP Logged
 
Rad
Radministrator
*****
Offline


Sufferin' succotash

Posts: 4090
Newport Beach, California


Back to top
Re: HTML compression and Apache mod_gzip
Reply #3 - Sep 28th, 2008 at 11:57pm
 
Thanks, bro.

I will watch for 500 server errors. Haven't noticed any errors yet. (mod_deflate compiled/enabled; as mod_gzip doesn't work with Apache 2.x)
 
WWW  
IP Logged
 
Rad
Radministrator
*****
Offline


Sufferin' succotash

Posts: 4090
Newport Beach, California


Back to top
Re: HTML compression and Apache mod_gzip
Reply #4 - Sep 29th, 2008 at 4:45pm
 
How can I (reliably) measure compression?

cuz, according to this page (which does identify HTTP compression), I am seeing *insignificant* compression, if that page is reporting accurately.

http://www.websiteoptimization.com/services/analyze/
 
WWW  
IP Logged
 
Nigel Bree
Ex Member




Back to top
Re: HTML compression and Apache mod_gzip
Reply #5 - Sep 29th, 2008 at 5:16pm
 
Rad wrote on Sep 29th, 2008 at 4:45pm:
How can I (reliably) measure compression? 

Don't try. It's a small and potentially useful part of the performance puzzle, but only a small part. And there are parts you should compress, but for most things it's not important.

I'd suggest you go with YSlow's analysis first, because that makes much smarter suggestions than that website does. Yahoo's guidelines are excellent, and read their blog entries too.

Compression is a distant second best behind really good cache control - setting Expires: headers appropriately, for instance. And where compression matters most are for page elements like external CSS and scripts that can block the page being rendered ahead until they are loaded, not for the content itself. You definitely should minify any Javascript and set it to be gzipped (after setting up Expires: headers, mind you) and that will typically be the only thing that really matters to the final end-user perception of page loading, because your main goal is to start the page rendering quickly. How things load after that isn't nearly as important.
 
 
IP Logged
 

Rad
Radministrator
*****
Offline


Sufferin' succotash

Posts: 4090
Newport Beach, California


Back to top
Re: HTML compression and Apache mod_gzip
Reply #6 - Sep 29th, 2008 at 7:46pm
 
Thanks Nigel. Trove of good info there. Reading now. Well written in simple terms. Installed YSlow (appropriately named).
 
WWW  
IP Logged
 
Page Index Toggle Pages: 1
Send Topic Print