The Red Hat Enterprise Linux 4 Update 6 kernel exhibits a large drop in sequential write performance primarily to a block device

Solution Verified - Updated -

Issue

  • This is a performance regression from RHEL4.5 in single stream sequential write workloads when using the default (CFQ) I/O scheduler.
  • The problem is restricted to performing I/O to a block device or an ext2 filesystem.
  • The problem is not present when using the ext3 file system.

For block device

Warning: this will overwrite any existing data on the block device:

RHEL 4.5
  # time dd if=/dev/zero of=/dev/sdb1 bs=1M count=3000
  3000+0 records in
  3000+0 records out

  real    0m36.969s
  user    0m0.013s
  sys     0m13.224s
RHEL 4.6
  # time dd if=/dev/zero of=/dev/sdb1 bs=1M count=3000
  3000+0 records in
  3000+0 records out

  real    1m6.073s
  user    0m0.014s
  sys     0m8.659s

For an ext2 filesystem

Warning: this will overwrite the file "testfile" in the working  directory, if it exists:

RHEL 4.5
  # time dd if=/dev/zero of=testfile bs=1M count=3000
  3000+0 records in
  3000+0 records out

  real    0m39.362s
  user    0m0.017s
  sys     0m9.689s
RHEL4.6
  # time dd if=/dev/zero of=testfile bs=1M count=3000
  3000+0 records in
  3000+0 records out

  real    0m49.116s
  user    0m0.011s
  sys     0m9.231s

Environment

  • Red Hat Enterprise Linux 4 Update 6, 7, 8 (all architectures).
  • Red Hat Enterprise Linux 4 Update 9 for x86.
  • A block device or an ext2 filesystem.

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
Close

Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.