This may seem a bit ‘back to basics’, but it’s a topic of discussions I’ve been having recently.
When presenting performance metrics, folks sometimes use mean, sometimes median, sometimes both, sometimes they include standard deviation, percentiles, etc. I’ve been looking for some concrete guidance on what metrics to use, and in what contexts.
There has been a good deal of work around defining guidelines, best practices, etc. for improving Web page performance (i.e. delivery speed). As Web page publishers have worked to implement these best practices, they inevitably reach a point where they can go no further to improve their own content. The long poles in the tent are outside their control, often in third-party content - in many cases, ads.
Ensuring the static objects from your Web site are cacheable is one of the best things you can do to improve performance.
I stumbled across this site the other day, with a tutorial on caching, and a tool to assess the cacheability of a site. Very nice.
AOL has released an open-source tool for measuring Web page performance called PageTest. It shares some similarities with Yahoo!’s YSlow, but works with Internet Explorer (6 and 7).
You can run PageTest as a browser plug-in, or use a hosted version at webpagetest.org (this is still being built out, so please be patient).
We’ve been using (what has evolved into) PageTest at AOL for years. Give it a try.
What keeps users coming back to your Web site? Compelling content? Fast Performance? Cool site design? The answer is “yes”.
Yahoo! released version 0.9 of YSlow recently. The release notes offer the highlights, but this blog entry by one of the developers provides more details.
I haven’t had a chance to spend much time with the new release, but it promises to address what I saw as the biggest shortcoming of prior releases - that it only crawled the DOM and didn’t capture network traffic. The new version promises to do both, providing a much more complete picture.
I was a bit disappointed to see that a bug I had reported in prior versions still remains - YSlow falsely identifies redirects. For example, today on www.aol.com there are 3 redirects, but YSlow identified 20 - the other 17 are listed as “redirects to <blank>” .
Nevertheless, a great tool.
I’ll be giving this presentation on Apdex at the CMG International Conference next week.
Ran across this old(er) article on techniques to improve Web site performance, written by Cal Henderson of Flickr/Yahoo.
Good stuff.
