Please remember to mark the question as Answered when you get the correct or best answer. Also, please award points to the person providing the best answer.
I am still battling with the Application owner as to the testing parameters they are using with iometer. The FC luns are performing as they want, but when we also add the SATA luns into the test we are filling up cach and then performance is affected everywhere. We believe their testing parameters are not set correctly and our local EMC rep and the application owner are meeting to discuss the iometer settings. The I/O footprint for the SATA luns is 4k block size 30% random writes times four servers. We may have to change the SATA from Raid 5 4+1 to Raid 1_0 4+4. Will update when the next test completes. All of the answers thus far have been helpful in narrowing in on a fix. Thank you
As an add, you can take your actual disk specs and throw them in WMarrow's calculator with data to match your iometer and host tests and determine approximatly what IOPS you should expect:
In IOmeter - you can set the IO size - not sure about the application - I just wanted to show the difference between 4KB and 32KB in the number of IO's.
I will ask them to change the I/O size in iometer before the next test. Unfortunately we do not have access to the tool or servers so I must rely on our Wintel team to make the changes, run the test and get the results that the Company “Autonomy” parses through to give me the results. I am going to pull SP collects and the .NAR file to see how the array performed during the test. Just not sure how to compare apples to apples between the iometer results and the .NAR file results.
easiest way is to look at the LUN under test, determine what the speed of the disks are (7200, 10K, 15K), how many disks are in the raid group (2+2 or 4+4 etc), then look at the type of test - is the test all reads or all writes. Then determine what the raid group can handle - see the Best Pratice guide in the "Size the Storage Requirements" section - there are a number of formula for determining the capactiy of the raid groups.
EMC CLARiiON Performance and Availability Release 30 Firmware Update Applied Best Practices.pdf
EMC CLARiiON Storage System Fundamentals for Performance and Availability