2014-06-24 10:11 AM
Here regarding throughput restriction (QoS), i am having one doubt. I want to clarify my doubt from below scenario.
Example, i am having four 2 TB SATA disks, each of which having 70 IOPS. Total 280 IOPS. All four data disks. If i create four 1 TB Volume (250 GB from each disks...) then "Does EACH volume will support maximum 280 IOPS ,OR, ALL FOUR volumes will share max 280 IOPS" ? .
I am asking this question because while creating volumes in our enviroment, we are applying qos policy group (2000 IOPS) on each volume then we were getting performance issue, i checked in command prompt (qos stat perf show) where i observed that applied qos policy group was giving maximum 2000 to 2500 IOPS total. If it is sharing the total throughput then what is the use of applying thoughput limit (IOPS) on each storage objects (Volumes, LUNs, Files).?? Please correct me if i am wrong.
2014-06-24 01:44 PM
All four volumes will share 280 IOPS. But remember that the SATA disk 70 IOPS you mentioned are 100% random IOPS. If you do sequential IOs you will get much higher number (and remember that OnTap translates random writes to sequential writes)