Multi-threaded concurrent direct IO is slower on EXT filesystems versus raw block devices

Solution Verified - Updated -

Issue

  • Multi-threaded IO utilizing O_DIRECT is slower on EXT2, EXT3, and EXT4 than it is on a block device
  • An example workload such as a restore from a Sybase backup where Sybase is configured to use direct IO is much slower when using an EXT filesystem versus a raw block device
  • The speed difference is as much as 200% slower on a filesystem versus raw block device

Environment

  • Red Hat Enterprise Linux 4
  • Red Hat Enterprise Linux 5
  • Red Hat Enterprise Linux 6
  • EXT2
  • EXT3
  • EXT4
  • Multi-threaded concurrent direct IO

Subscriber exclusive content

A Red Hat subscription provides unlimited access to our knowledgebase of over 48,000 articles and solutions.

Current Customers and Partners

Log in for full access

Log In