Announcement

Collapse
No announcement yet.

Trying to explain some unexpected results

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Trying to explain some unexpected results

    I have a database server that is being replaced and just for the hell of it ran phoronix on the old one to compare to the new one and don't see the increase in the performance one might expect. The system details and test results are in the attached image, but highlights are:

    Old-
    Xeon E5640 @2.67GHz X 4 cores
    384GB RAM
    RHEL 5.5 (2.6.18 kernel)

    New-
    Xeon E5-4650 @2.7GHz X 64 cores
    1TB RAM
    OEL 6.4 (2.6.39 kernel)

    In particular, the RAMspeed tests indicate degraded performance on the new system, and even in the tests that are better, the results aren't exponentially better. The old system is booting from local disk, db is on SAN. New system is booting from SAN, db on san, /tmp is tmpfs in RAM.

    I know the real test will be when the db is loaded and tested, but just basic benchmark tests are not showing spectacular improvements.


  • #2
    Some tests locked at single core? Core migration would explain further slowdowns of such tests. Try running htop or slowly increasing multiple test instances and then sum up all ratings.

    Comment


    • #3
      update

      I'm working with Cisco now to optimize/tune the OS now. I'll update with findings next week.

      Comment

      Working...
      X