Bug 754960

Summary: MD RAID performance regression
Product: Red Hat Enterprise Linux 6 Reporter: Eugene Apollonsky <e.apollonsky>
Component: kernelAssignee: Jes Sorensen <Jes.Sorensen>
Status: CLOSED INSUFFICIENT_DATA QA Contact: Red Hat Kernel QE team <kernel-qe>
Severity: medium Docs Contact:
Priority: unspecified    
Version: 6.1CC: dledford, Jes.Sorensen
Target Milestone: rc   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2011-11-18 14:16:09 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Attachments:
Description Flags
Performance measurements none

Description Eugene Apollonsky 2011-11-18 11:50:05 UTC 
Description of problem:
We have strange performance regression on RHEL6.1 versus RHEL5.2. See attachment for more details.

Actual results:
Average value and standard deviation of data rate on RHEL6.1 are worse than on RHEL5.2.

Version-Release number of selected component (if applicable):
Standard RHEL6.1 distribution.
Comment 1 Eugene Apollonsky 2011-11-18 11:51:36 UTC 
Created attachment 534377 [details]
Performance measurements
Comment 3 Jes Sorensen 2011-11-18 14:16:09 UTC 
Eugene,

dd from /dev/zero is really not a good way to test I/O performance, it is
impacted on how /dev/zero is generating data, it is single threaded etc.

Please try and benchmark it using a real disk benchmark tool like iozone,
or fio, etc.

In addition you do not provide any data about the memory of the system, number
of processors, how the disks are configured, etc.

Note that your RHEL5.2 benchmark numbers are very suspicious as well, the
fact that halt the numbers are exactly 2048 with no variation looks very
bizarre.

Best regards,
Jes