Optimizing Web Pages

With the new Web site I’m developing drawing nearer to its public debut, and with the entire backend being written from scratch by me, I’ve become concerned with optimizing the output as much as possible.

While searching for some resources the other night (I was mainly looking for a Firefox add-on that would display a page’s load time, the way Netscape used to do – the only one I found has not been updated to work with FF3, yet), I came across two interesting resources.

WebSlug

WebSlug.info is a Web site that measures the amount of time it takes to load a site. Some really useful aspects of the site, in addition to the fact that it measures the load time, are the following:

  • WebSlug allows you to test two sites against each other. If you’re working on a new Web site (as I am), you can actually compare your old site against your new site. Or, if you’re feeling really froggy, you can try to compare your site to a really fast site like Google or neotelos.com (the top-ranked site in WebSlug’s database at the time I’m writing this article, with an average load time of 0.15 seconds).
  • WebSlug keeps track of every test that’s run through its interface, allowing you to track the average load time over a series of tests. As mentioned above, neotelos.com lists an average of 0.15 seconds per load, after being tested 256 times. Google is the second-ranked site in WebSlug’s database, with an average load time of 1.48 seconds over 746 tests.

The odd thing about WebSlug’s rankings is the fact that they seem to be based on some sort of “score” combining the average load time and the number of recorded tests. Therefore, although Google has the fifteenth fastest load time on WebSlug’s “top 20 sites” page, it’s ranked second, because it’s been tested more than any other page on the list.

WebSlug also allows you to view the top 20 and bottom 20 sites tested with its interface.

The main problem I’ve found with WebSlug is the fact that it uses your connection and browser to run the test. Therefore, you might not always get an accurate snap shot of how your sites perform in comparison to each other. For most people, the difference would probably never be noticeable.

For me, however, the tests are completely inaccurate when I run them from work. Because I have a direct connection to our Web server, meaning that our in-house Web site loads immediately every time, it offsets the results. Although our current Web site took an average of about 15-20 seconds to load from home, it took less than half a second to load at work. Therefore, this test is really only reliable when running it on a remote Web site.

The fact that the test is run through your browser could also potentially offset the results. Although most of us may not ever notice a major difference in the speed of our browsers, there are differences. For instance, the default installation of Opera is apparently considerably faster than the default installation of Firefox of IE.

While comparing sites would most likely be accurate, the records stored in the database cannot really be considered as such, being that one site may be tested using nothing but IE on a dialup connection, while another site might be consistently tested using Opera on a T1 connection.

WebSlug does keep some analytical information on the tests that are run, allowing you to view a breakdown of which browsers and which operating systems were used (though the connection type and speed are not tracked), so you can try to break things down that way.

Die.Net

I also found an interesting blog article at a Web site called die.net, detailing some intriguing research and some nice tips on optimizing page loads.

One Response