Short Story about IO-Errors
Hardware failure is not an exception, it is a normal process. I guess most of…Read More
There are plenty of benchmarking tools available but storage professionals mostly use IOmeter. Unfortunately IOmeter is a little tricky to use. You really need to read the user manual first. I have frequently seen users trying to run IOmeter tests without success. Being human, most of us hate to read the manual and with IOmeter this can lead to problems. I hope this short post will help you to get the wanted results. First off, you need to know that IOmeter recognizes 2 different volume types:
• un-partitioned disks (blue icon disk), or
• formatted disks (yellow icon with a red slash though it)
With un-partitioned disks you can start the test at once, but to run with formatted disks you need a test file. The test file must be placed in the root directory and named: iobw.tst. By default IOmeter will create the test file if not found. The problem is that nowadays volumes are very big, and IOmeter runs very slowly. It’s much faster to create the test file using the TestFileCreator.exe from Open-E. Please run it in order to create iobw.tst with any size you desire. You can run: TestFileCreator.exe 100G in order to create a file of exactly 100GB.
To find out your storage performance there are a few typical test configurations you may want to run. Here are some example results of a FC Volume created on DSS V6 with FC HBA dual 4Gb and MPIO.
So, if your goal is to obtain the maximum MB/sec please create a test set up with 2-4 workers, using a block size of 256k and 100% sequential read or write. In our case the sequential read shows the best result: 772 MB/sec! Please make your settings very carefully and make sure all workers use the same test configuration! If you forget to add your [256kB, 100% sequential read] configuration to every worker you will be surprised with very low results because the default test settings use 2kB block and mixed random/sequential and read/write pattern. So instead of the desired 772MB/sec you might see 100 times less, i.e. ~ 7MB/sec.
ErrolDCDecember 01, 12 2010 03:24:40
IOmeter is such a great tool for understanding storage performance. One of things I’d like to see in future open-E releases is the ability to test performance on the Storage Server side. Dynamo is the server component of IOMeter for Linux. You can integrate it and add some basic controls in the UI so that someone from a Windows system can connect to Dynamo remotely and run tests. This gives the most accurate picture of storage performance. Many people make the mistake of trying to test performance on the initiator side without truly understanding what the underlying storage is capable of delivering.
todd maxwellFebruary 01, 02 2011 01:15:11
We have a small update that can provide some performance on Reads and Writes for Units that have not been formatted – email me at pre-sales@open-com and I will provide this for you also for the NICs as well (you will need another DSS V6 to perform this).
David WNovember 01, 11 2012 12:46:13
The problem with Iometer is the results output depend on the size of the iobw.tst file. If you create a small file you get very good results because you are running in cache. If you create a large .tst file, say over 2GB you get very different results.
Use with care and use consistently. Unfortunately people are using this all over the internet without clarifying how they have setup the tests.
PrasadMarch 05, 03 2013 07:39:09
Is there any CLI command to stop the running IOmeter execution without using user interface stop button.
Kasia KolodziejMarch 06, 03 2013 10:18:08
Hey Prasad! The IOmeter tool is a separate software and there is no option to use the CLI command, since it is not included in our product. I would suggest asking the IOmeter’s support team if they have an option like this. You can find more information about the IOmeter here: http://www.iometer.org/doc/screenshots.html
QuantumMarch 19, 03 2013 08:25:28
So I am running IOMeter with dual 8GB FC ports from QLogic, when doing Writes only, I notice that IOMeter is only using 1 FC port out of the 2 (is this by design or there is a configuration file somewhere that we need to mod for 2 ports FC to work). For Reads, IOMeter does use both FC ports so aggregate is about 1.6GB/Sec compare to 750MB/Sec for Writes. Anybody has seen this issue on their end?
Kasia KolodziejMarch 25, 03 2013 01:26:17
It might be that you will have to configure MPIO on the FC initiator side. An explanation of how to do this is available in our knowledgebase:
(the part describing configuration of dm-multipath is the same for FC and iSCSI)
Hope this helps!
YinJuly 20, 07 2015 06:34:19
need some help please .
i need to test 50Gb Big files * 4 copies in parallel repeating (copy, delete, copy) for 1/2 hour.
Can any one tell me how to configure ?
karolina.pletniaJuly 22, 07 2015 10:22:42
Unfortunately we are not able to help you by answering your comment. Please create a ticket with our Technical Support team. They will send you a personalized IO meter profile. In order to create a support ticket, please go to your Open-E account here: https://www.open-e.com/portal/login/?next=/userportal/
Michael NJuly 29, 07 2015 09:48:03
we got a win2k12 server with 2 8Gb HBAs cnnected to a HDS G1000. the win-host is mapping virtualized volumes from the G1000 configured as raw-device (not initialized). when we’re starting an iometer-test for example with 60%read 40%write we only see read-transfers on G1000. when starting the same profile with 100%write the IOPS on initialized partitions are crashing down to 40-50 IOPS. is this problem known ?
Kasia KolodziejAugust 04, 08 2015 07:13:00
Hey Micheal. Please make sure that in Windows Disk Management the disk is set to online and initialized. Hope this helps!
Jarred WaltonAugust 10, 08 2015 09:33:47
I’m not sure of the cause, but it seems the first test run with IOmeter after using TestFileCreator will give garbage results. For example, on a fast Intel SSD 750 1.2TB drive, the first test I ran on a 128GB test file was a 128K sequential 100% read for six minutes, with a 60 second warmup. The drive supposedly scored 220,000MB/s, or about 10X higher than it should. All the other test results were basically as expected, and when I reran the 128K Seq 100% read again, the result the second time was correct (2272MB/s). I’ve noticed this on basically every drive I’ve tested.
NaidJanuary 09, 01 2016 05:02:27
Hi, how long it takes at Run Time to get a valid result from SSD, especially the result maximum MB/sec. Thanks before. 🙂
karolina.pletniaJanuary 14, 01 2016 01:23:58
Hello Naid! In Open-E DSS V7 IOmeter should show valid results right away. In Open-E JovianDSS in turn, the read results may take a while, even up to a few hours in some cases, before read cache learns the data.
ErnieJanuary 27, 01 2017 03:07:39
Your suggested test is for 100% Sequential Reads. Note that most modern storage systems have a “look ahead” feature that preloads the next block of storage in to RAM. This would return a large number that might not necessarily represent real world more random type access.
DanielleMarch 10, 03 2018 12:46:06
How do you use the TestFileCreator software? I can’t find the iobw.tst file its creating. Also I’m not sure if this is normal but nothing pops up when I open it.
ChrisApril 05, 04 2019 10:12:57
Is there a way to get IOMeter to stop automatically after a single sequential pass on a drive?
Whenever I run the benchmark, it just continues to run!
Karolina NapierajApril 23, 04 2019 07:55:40
Hi Chris! IOMeter is an external tool, please refer to the IOMeter PDF User Guide available here: http://csis.pace.edu/~lombardi/sciences/computer/systems/windows/docs/iometer.pdf or contact IOmeter developers.
As for the User Guide, check page 15: “5.10 Test Setup Tab — Introduction” you will find useful information.
Hope this helps!
Bernd LeuenrothJuly 03, 07 2021 06:16:44
When trying to orchestrate a test in which an IOmeter test run takes place simultaneously from several
Windows10 workstations on a network drive, I found that IOmeter version 1.1.0 is not suitable for Windows10
task scheduler when the task is bound to run in the background with “highest priorities” and without a
Instead of the expected results in CSV files, crashes occur, which generate the following events:
Event ID: 1000
Source: Application Error
Faulting application name: IOmeter.exe, Version: 18.104.22.168
Exception code: 0xc0000409
Event ID: 1000
Source: Application Error
Faulting application name: Dynamo.exe, Version: 22.214.171.124
Exception code: 0xc0000005
In the zip archive
you can find a solution to the problem described above.
The following files are included:
– the executable files IOmeterX.exe and the associated DynamoX.exe
– README.DE.txt in German
– README.EN.txt in English
– IOmeterX.patch source code patch file for the changes between iometer-1.1.0-src.tar and IOmeterX v1.1.1
optional: Apply the source code patch on a Linux machine: cd iometer-1.1.0 ; patch -p0 -i IOmeterX.patch
Extract *.exe into the same directory where IOmeter.exe was installed.
Then call IOmeterX.exe instead of IOmeter.exe.
Shubhadip PalJuly 14, 07 2021 07:49:30
Is there any option in IOmeter to mention cache hit % ?
Is there any option is IOmeter to mention specific iorate to see the response time for that iops?