Hi, Not sure if you have been sorted. Only came across your post now. Based on your setup you only have about 280 RAW IOPS available. Let me explain, a 7200RPM disk gets around 80 IOPS. You have 14 of those ( 14 * 80 = 1120), divide that by 4 because of your RAID50 penalty and you get to 280 usable IOPS on your array. Of course cache helps but only depending how much your workload is requesting. Let's say your SQL environment is requesting 2000 IOPS, your disk wont keep up with your cache. Hence your cache will run full and you will start queuing on disk. I would suggest looking at your avg Disk Write/sec and avg Disk Read/sec, these should be below 20ms, or < 0.0020. If above 20ms you have disk latency, start looking at your Disk Read/sec and Disk Write/sec to see what your R/W balance is and spec your disk accordingly.
↧