Home / server / Questions / 121748
Accepted
PJ.
Asked: 2010-03-12 17:06:22 +0800 CST

How GZipped contents are transfered on the web?

  • 772

I heard that static contents like CSS and JavaScript can be better delivered in GZip format. And Content Developer Network (CDN) always does so.

However I don't understand how the format works. First when I tried making a gzipped file via command-line. The file extension is .gz. This is different from .css and .js. How do browsers recognize which file is gzipped or not.

Second, how browsers "decompress" files? I dragged my index.html.gz onto my browsers. But no one worked.

  • How do such gzipped work in the real world?
  • What do I need to do if I want to serve CSS/JavaScript using Gzipped format.
hosting web-server gzip

2 Answers

  1. Best Answer

    The gzipping you're talking about is invisible to the end user, and for that matter, to you as a developer. It's taken care of by your web server, and the client's web browser.

    For IIS, you need to enable the GZip ISAPI extension (check around on this site for a how-to), and for Apache it's mod_deflate.

    When the web browser has completed its request, but before sending it out into the wide world, it will gzip the content and put a special header in, so that the browser knows the content is compressed.

    When the web browser receives the response, it then unzips the data, and reads the file. This is completely transparent to the end user.

    You usually only want to use GZip on text files. There's no point in using it on images, as images are already heavilly compressed.

    I noticed a drop from 90Kb page size (load time of approx 1 second) to about a 5Kb page size (load time, .4 seconds) when enabling GZip on all static content (CSS, Javascript, etc).

    What you would find if you watched the packets with Wireshark and re-assembled them, you would request /index.html, but the webserver would fetch /index.html, GZip it into /index.html.gz and send it off. Yhe web browser knows this and unzips it so that it's back to /index.html, so when you go to View Source you see /index.html, not /index.html.gz

    • 4

  2. Generally Gzip compression of static content from a web server is done via Apache mod_deflate

    http://httpd.apache.org/docs/2.0/mod/mod_deflate.html

    This is usually only done with text files in the formats of text/plain, text/html and text/xml, it can be done with other plain text content types (css, js etc) but that can cause issues with early versions of internet explorer. (edit:aparrently v6 and below)

    Images should be compressed already in what ever format they are stored in, so there should be no additional (ie mod_deflate) compression applied to those.

    • 3