You are previewing High Performance Web Sites.

High Performance Web Sites

Cover of High Performance Web Sites by Steve Souders Published by O'Reilly Media, Inc.
  1. High Performance Web Sites
    1. SPECIAL OFFER: Upgrade this ebook with O’Reilly
    2. Praise for High Performance Web Sites
    3. Foreword
    4. Preface
      1. How This Book Is Organized
      2. Conventions Used in This Book
      3. Code Examples
      4. Comments and Questions
      5. Safari® Books Online
      6. Acknowledgments
    5. 1. The Importance of Frontend Performance
      1. Tracking Web Page Performance
      2. Where Does the Time Go?
      3. The Performance Golden Rule
    6. 2. HTTP Overview
      1. Compression
      2. Conditional GET Requests
      3. Expires
      4. Keep-Alive
      5. There's More
    7. 3. Rule 1: Make Fewer HTTP Requests
      1. Image Maps
      2. CSS Sprites
      3. Inline Images
      4. Combined Scripts and Stylesheets
      5. Conclusion
    8. 4. Rule 2: Use a Content Delivery Network
      1. Content Delivery Networks
      2. The Savings
    9. 5. Rule 3: Add an Expires Header
      1. Expires Header
      2. Max-Age and mod_expires
      3. Empty Cache vs. Primed Cache
      4. More Than Just Images
      5. Revving Filenames
      6. Examples
    10. 6. Rule 4: Gzip Components
      1. How Compression Works
      2. What to Compress
      3. The Savings
      4. Configuration
      5. Proxy Caching
      6. Edge Cases
      7. Gzip in Action
    11. 7. Rule 5: Put Stylesheets at the Top
      1. Progressive Rendering
      2. sleep.cgi
      3. Blank White Screen
      4. Flash of Unstyled Content
      5. What's a Frontend Engineer to Do?
    12. 8. Rule 6: Put Scripts at the Bottom
      1. Problems with Scripts
      2. Parallel Downloads
      3. Scripts Block Downloads
      4. Worst Case: Scripts at the Top
      5. Best Case: Scripts at the Bottom
      6. Putting It in Perspective
    13. 9. Rule 7: Avoid CSS Expressions
      1. Updating Expressions
      2. Working Around the Problem
      3. Conclusion
    14. 10. Rule 8: Make JavaScript and CSS External
      1. Inline vs. External
      2. Typical Results in the Field
      3. Home Pages
      4. The Best of Both Worlds
    15. 11. Rule 9: Reduce DNS Lookups
      1. DNS Caching and TTLs
      2. The Browser's Perspective
      3. Reducing DNS Lookups
    16. 12. Rule 10: Minify JavaScript
      1. Minification
      2. Obfuscation
      3. The Savings
      4. Examples
      5. Icing on the Cake
    17. 13. Rule 11: Avoid Redirects
      1. Types of Redirects
      2. How Redirects Hurt Performance
      3. Alternatives to Redirects
    18. 14. Rule 12: Remove Duplicate Scripts
      1. Duplicate Scripts—They Happen
      2. Duplicate Scripts Hurt Performance
      3. Avoiding Duplicate Scripts
    19. 15. Rule 13: Configure ETags
      1. What's an ETag?
      2. The Problem with ETags
      3. ETags: Use 'Em or Lose 'Em
      4. ETags in the Real World
    20. 16. Rule 14: Make Ajax Cacheable
      1. Web 2.0, DHTML, and Ajax
      2. Asynchronous = Instantaneous?
      3. Optimizing Ajax Requests
      4. Caching Ajax in the Real World
    21. 17. Deconstructing 10 Top Sites
      1. Page Weight, Response Time, YSlow Grade
      2. How the Tests Were Done
      3. Amazon
      4. AOL
      5. CNN
      6. eBay
      7. Google
      8. MSN
      9. MySpace
      10. Wikipedia
      11. Yahoo!
      12. YouTube
    22. Index
    23. About the Author
    24. Colophon
    25. SPECIAL OFFER: Upgrade this ebook with O’Reilly
O'Reilly logo

AOL

Figure 17-6. http://www.aol.com

The HTTP requests for AOL (http://www.aol.com) show a high degree of parallelization of downloads in the first half, but in the second half, the HTTP requests are made sequentially (see Figure 17-7). In turn, the page load time is increased. There are two interesting implementation details here: downgrading to HTTP/1.0 and multiple scripts.

AOL HTTP requests

Figure 17-7. AOL HTTP requests

In the first half, where there is greater parallelization, the responses have been downgraded from HTTP/1.1 to HTTP/1.0. I discovered this by looking at the HTTP headers where the request method specifies HTTP/1.1, whereas the response states HTTP/1.0.

GET /_media/aolp_v21/bctrl.gif HTTP/1.1
Host: www.aolcdn.com
HTTP/1.0 200 OK

For HTTP/1.0, the specification recommends up to four parallel downloads per hostname, versus HTTP/1.1's guideline of two per hostname. Greater parallelization is achieved as a result of the web server downgrading the HTTP version in the response.

Typically, I've seen this result from outdated server configurations, but it's also possible that it's done intentionally to increase the amount of parallel downloading. At Yahoo!, we tested this, but determined that HTTP/1.1 had better overall performance because it supports persistent connections by default (see the section ...

The best content for your career. Discover unlimited learning on demand for around $1/day.