Solved

LVM Large File Performance Test

Posted on 2010-08-18
3
684 Views
Last Modified: 2012-05-10
I ran tests to compare disk performance of LVM versus normal ext3 partitions
on a hardware Raid 10 consisting of six 600G 10k physical disks.  According to my
google search, I was expecting the performance to be about the same or a
tad bit slower for an LVM configuration.  

The LVM consisted of 6 logical volumes on one volume group using the whole Raid 10.
The non-LVM consisted of 6 ext3 partitions on the Raid 10.

Can anyone explain why the LVM performed considerably better?
I was seeing a peek of 400k sectors/second on the LVM and
a peek of 234k sectors/second on the non-LVM.  

Test Description
================
a) Test consists of concurrent I/O of 24 large 2G files
b) Memory cache is cleared before each test
c) Disk cache is allowed to quiesce before each test

=====================
Large File Write Test
=====================
229 seconds - Dell R710 (LVM)
389 seconds - Dell R710 (no LVM)

====================
Large File Read Test
====================
296 seconds - Dell R710 (LVM)
521 seconds - Dell R710 (no LVM)

The write test consisted of writing 24 2G files using the dd command.
The read test consisted of copying those 24 2G files to /dev/null.
0
Comment
Question by:otwayc
  • 2
3 Comments
 
LVL 4

Accepted Solution

by:
DeadRatRacer earned 500 total points
ID: 33471686
First thought is that under the LVM the logical partition sectors were correctly aligned against the disk's physical layout and the ext3 partition was not -  just a guess though.

Lots of articles, here is but one on the topic
http://blogs.sun.com/dlutz/entry/partition_alignment_guidelines_for_unified
0
 
LVL 4

Expert Comment

by:DeadRatRacer
ID: 33471695
More obscure post, but showing a real-world 6x increase in speed under ext3 after proper alignment

http://aulix.com/partition-alignment
0
 

Author Comment

by:otwayc
ID: 33493405
Thanks.  Yes, it apparently is very sensitive to the parameters.  The Oracle OEL installation screens have a default LVM chunk size of 32M which performed quite a bit worse than the OEL Linux PVCREATE/VGCREATE default value of 4M (for this particular test).  I noticed the LVM configuration seemed to maintain a more even balance of I/O across the physical disks than a basic EXT3 configuration.  This could very well be due to the parameters as you suggest.
0

Featured Post

PRTG Network Monitor: Intuitive Network Monitoring

Network Monitoring is essential to ensure that computer systems and network devices are running. Use PRTG to monitor LANs, servers, websites, applications and devices, bandwidth, virtual environments, remote systems, IoT, and many more. PRTG is easy to set up & use.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

Suggested Solutions

Solid State (Hard) Drives aka SSD began to evolve in the computer industry recently. As the name suggests, there are no moving parts in the drives. The drive uses microchip memory store the data, as opposed to the spinning disks of a traditional HDD…
Ever notice how you can't use a new drive in Windows without having Windows assigning a Disk Signature?  Ever have a signature collision problem (especially with Virtual Machines?)  This article is intended to help you understand what's going on and…
This video Micro Tutorial explains how to clone a hard drive using a commercial software product for Windows systems called Casper from Future Systems Solutions (FSS). Cloning makes an exact, complete copy of one hard disk drive (HDD) onto another d…
Finds all prime numbers in a range requested and places them in a public primes() array. I've demostrated a template size of 30 (2 * 3 * 5) but larger templates can be built such 210  (2 * 3 * 5 * 7) or 2310  (2 * 3 * 5 * 7 * 11). The larger templa…

829 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question