Scaling WordPress
4 likes1,866 views
This document discusses optimizing a WordPress site to handle high traffic loads. It provides tips for caching at various levels (opcode, object, page, fragment) using tools like Nginx, PHP-FPM, memcached. It also recommends using a CDN, handling traffic variability, and scaling to multiple application servers with a load balancer. Benchmark results show performance improving from serving 50k pages/day to over 8k requests/second after these optimizations.
![$ ab -c 350 -n 5000 http://havebabyneedstuff.com/
Result?
Requests per second: 8114.53 [#/sec] (mean)
95% of requests served within 40ms
99% of requests served within 82ms](https://image.slidesharecdn.com/scalingwordpress-121201080057-phpapp01/85/Scaling-WordPress-25-320.jpg)
Scaling WordPress
- 1. Mark Jaquith Making WordPress “JAKE-with” @markjaquith Go “Vroom” mark@jaquith.me markjaquith.com
- 2. Survey
- 3. MATH TIME 50k pages/day = 0.58 pages/second Uh... that’s it?
- 4. Traffic isn’t steady
- 5. Speed Ma ers
- 7. Apache
- 8. nginx
- 9. proxy to Apache or PHP-FPM
- 10. PHP Opcode Cache
- 11. WP Object Cache
- 12. HTML Output Cache Batcache || W3 Total Cache || WP Super Cache
- 13. Nginx Microcache
- 14. Fragment Caching if ( $html = get_transient( 'expensive-html-chunk' ) ) { echo $html; } else { $html = dynamically_generate_foo(); set_transient( 'expensive-html-chunk', $html, 3600 ); echo $html; }
- 15. TLC Transients echo tlc_transient( 'example-feed' ) ->updates_with( 'my_callback' ) ->expires_in( 300 ) ->background_only() ->get();
- 17. CDN — with — Origin Pull
- 18. Guideline: 1000ms+ — Really slow 350–999ms — Slow 250–349ms — Okay 100–249ms — Fast 0–99ms — ZOMG FAST
- 19. The Optimization Cycle
- 20. Dead-simple bo leneck identification <!--timer: <?php timer_stop(true) ?>-->
- 22. Nginx caching
- 24. Result?
- 25. $ ab -c 350 -n 5000 http://havebabyneedstuff.com/ Result? Requests per second: 8114.53 [#/sec] (mean) 95% of requests served within 40ms 99% of requests served within 82ms
- 26. Multiple Servers
- 27. Load Balancing
- 28. Sharing Uploads
- 29. Memcached
- 30. Rinse & Repeat
- 31. HyperDB
- 32. Mark Jaquith “JAKE-with” Thanks! @markjaquith mark@jaquith.me markjaquith.com