Linux: Random File I/O Regressions In 2.6

A recent discussion on the lkml looked into a reproducable random file I/O regression in 2.6 compared to the 2.4 kernel. Alexey Kopytov posted the benchmark results, attempting to simulate the workload of a database under intensive load. The tests were tried with all I/O schedulers, including the anticipatory, deadline and CFQ, and in all cases 2.4 outperformed 2.6. Read the report at KernelTrap.

43 Comments

  1. 2004-05-04 7:27 pm
  2. 2004-05-04 8:47 pm
  3. 2004-05-04 9:12 pm
  4. 2004-05-04 9:37 pm
  5. 2004-05-04 10:15 pm
  6. 2004-05-04 10:34 pm
  7. 2004-05-04 11:28 pm
  8. 2004-05-04 11:39 pm
  9. 2004-05-04 11:40 pm
  10. 2004-05-04 11:46 pm
  11. 2004-05-05 12:05 am
  12. 2004-05-05 12:40 am
  13. 2004-05-05 12:56 am
  14. 2004-05-05 1:20 am
  15. 2004-05-05 1:26 am
  16. 2004-05-05 2:34 am
  17. 2004-05-05 3:04 am
  18. 2004-05-05 3:13 am
  19. 2004-05-05 3:31 am
  20. 2004-05-05 3:31 am
  21. 2004-05-05 3:40 am
  22. 2004-05-05 3:42 am
  23. 2004-05-05 3:42 am
  24. 2004-05-05 3:45 am
  25. 2004-05-05 3:50 am
  26. 2004-05-05 3:53 am
  27. 2004-05-05 3:54 am
  28. 2004-05-05 3:55 am
  29. 2004-05-05 3:56 am
  30. 2004-05-05 3:56 am
  31. 2004-05-05 4:00 am
  32. 2004-05-05 4:00 am
  33. 2004-05-05 4:04 am
  34. 2004-05-05 4:09 am
  35. 2004-05-05 4:10 am
  36. 2004-05-05 4:15 am
  37. 2004-05-05 4:15 am
  38. 2004-05-05 4:17 am
  39. 2004-05-05 4:18 am
  40. 2004-05-05 4:19 am
  41. 2004-05-05 4:23 am
  42. 2004-05-05 4:41 am
  43. 2004-05-05 7:54 am