I need to answer some questions to a customer for VNX 5500 storage regarding IOPS received in a given RAID Group and impact of cache on it. Ex.: If there is a RAID Group of RAID 5 8D+1P disks, what kind of IOPS can we get? How would cache add to getting these IOPS? IF the complete storage is populated with these RAID groups, what is the kind of IOPS the storage would provide?
What addiotnal benefit would FastCache provide on this?
The USPEED tools are very useful. However, they require detailed information to use them effectively. From the level of the OP's questions, I suspect s/he does not have that level-of-detail at this time.
I believe Wei Chen's suggestion to be a better one. Reviewing Chapter 5, 'Storage System Sizing and Performance Planning', of EMC Unified Best Practices and for Performance and Availability: Common Platform and Block O.E. 31.5
would provide an overview of the process and the information needed to provision storage. Most users find the different sizing examples in that section to be very informative. The OP may even be motivated to do some 'back-of-envelope' spreadsheet modeling, using the examples in that section in preparation for engaging a USPEEDer.
chenw87
43 Posts
222
0
Posted April 19th, 2012 12:00
It depends on the drive type and the type of I/O.
For example, with small-block (16KB or less) random I/O, you can expect:
SSD - 3500 IOPS
SAS 15K RPM - 180 IOPS
SAS 10K RPM - 150 IOPS
NL-SAS 7.2K RPM - 90 IOPS
Refer to the Best Practices for Performance and Availability white paper for more information:
https://powerlink.emc.com/nsepn/webapps/btg548664833igtcuup4826/km/live1/en_US/Offering_Technical/White_Paper/h8268_VNX_Block_best_practices.pdf