Everything old is new again! Actually really interesting though-- I think it's great to get back to the fundamentals, and even better to hear how much it can really help.

The book looks great, reading it on Safari now.

Are we gonna get the rest of the book for free over here, or is this a teaser?

I doubt we'll get the whole thing but I read as much as I could on Safari, and hopefully he won't mind me sharing the companion website, it has all 14 rules but not all have examples yet.

Still great. This is a must buy as soon as it's ready.

http://stevesouders.com/examples/rules.php

The book looks great. I like it.

Is there a tool that analyses how the browser downloads the different elements on the page and gives something like the chart you posted?

You can get charts like this from IBM Page Detailer: http://alphaworks.ibm.com/tech/pagedetailer

I totally agree that reducing web page weight is a good idea. One thing to be aware of is that the user doesn't have to wait for all the files to load before the page is rendered. AFAIK the files are loaded in the order they are specified in the HTML doc, so once the HTML and CSS has loaded, the user will see the page with empty spaces where the images and swfs will be. That's why it seems odd on your graph that the stylesheet is loaded at the end.

Not everyone puts their stylesheets in the head of the HTML document. They should, and that's the topic of Rule 5. In the example above, the stylesheet and script at the end are most likely downloaded via DHTML in the page's onload handler. Scripts also delay the rendering of pages, as shown in this example: http://stevesouders.com/examples/move-scripts.php

Are you sure that the 'Use a CDN' item in the performance tab is working? I am using a CDN, and it's telling me that I'm not.

Fiddler2's new Timeline extension (http://www.fiddler2.com/Fiddler2/extensions.asp#Timeline) shows charts like this as well.

You can add your CDN to the YSlow configuration settings (see http://developer.yahoo.com/yslow/faq.html#faq_cdn).

Yes, Fiddler 2 has great HTTP request timeline charts.

pingdom has a web-based tool to make these charts http://tools.pingdom.com/fpt

Steve Souders has done a great job putting together these recommendations, and I applaud his book, but am I the only one that thinks Yahoo! happens to have one of the slowest and most bloated page loads around? It's kind of disappointing to realize that they've already squeezed and optimized these pages and they are still slow loaders! The culprit is almost always fat flash files for ads and video. Nothing new here, but interestingly, I don't see flash files listed on the chart above. Perhap they are lumped in with images. While Yahoo! uses mostly open source (php, etc.) Flash is proprietary and Adobe has done next to nothing to open it up compared to other internet technologies - even compared to other proprietary tech. That seems to me to be the biggest opportunity for page load improvement - open up the flash tech so we can better manage it at the developer and user levels.

I'm all for making Yahoo!'s pages faster, but I will point out that across the ten top sites in the U.S., Yahoo! is in the middle. Again, I'd still like to see things faster and we continue to work (hard!) on making that happen.

Your comment on ads is true. Third party ads have become the #1 slowdown on Yahoo! pages. I've started discussions with the IAB (Interactive Advertising Bureau) to setup a performance working group to get industry-wide support for improving the performance of ads. If you see bad ads, complain! Don't buy that brand. Complain to the publisher (like Yahoo!). This feedback will encourage advertisers to improve their performance practices so they don't drive away consumers.

We are starting to dig into Flash performance. The Flash group at Yahoo! has built some cool tools to analyze Flash performance. But it's going to take awhile to gather the best practices and make the tools public. We'll share what we have when we have it.

Steve, colleague saw your presentation at Future of Web Apps in London and suggested I take a look at what your group is doing. What are your feelings about hardware Application Front End appliances like F5's BIG-IP and Citrix's NetScaler? Think they are overkill if one could "fix" front end issues in software?

These hardware appliances are generally focused on delivery of the HTML document. Since that's less than 10-20% of the overall user experience, the higher priority is focusing on the other 80-90% of the time. So these are not a first priority when it comes to optimizing the user response time. If your focus is to improve your _back end_ performance (e.g., requests/second), these can play a role.

Steve, do you think there will eventually be a guideline involving cookies based off of the "When the Cookie Crumbles" article? Perhaps something YSlow could check?

If images are on the same server as a cookie it gets returned in the HTTP request. If you have a site with a lot of new images, such as e-commerce, a lot of savings could be made. For example, on a search results page or a page that features new products with unique images. Since a cookie can be 4-5kb it seems that a large cookie could have a pretty sizable performance impact.

Hi, Rob. Yes, I'm baking 5-6 new rules now that address cookies and cookie-free domains. That definitely helps.

Hi!
I look at the chart that illustrates HTTP traffic and I am wondering about how did you get it. Did you use any measuring tool that analizes and provides this information on the fly or add some "smart" javascript measuring code to the page? Thanks.

See the comment from May 22.

Thanks ... Dude...

Hi, Rob. Yes, I'm baking 5-6 new rules now that address cookies and cookie-free domains. That definitely helps.

hey there,
Are you guys looking for charting component if yes then chk this out visifire i really enjoyed drawing chart with various animations

First, thanks for Your helpful work, Steve!

Take a look at http://site-perf.com/, may be You'll find it useful.
It's another online site download measurement tool, like pingdom.com, but looks more realistic.

Steve, just wondering how the Flash performance work is coming along.

Thank you very much for this useful article and the comments.

Hi,Its really nice post.I totally agree that reducing web page weight is a good idea. One thing to be aware of is that the user doesn't have to wait for all the files to load before the page is rendered.