Release Notes

39 Analyzing Dell PS Series Storage with SAN Headquarters | TR1050
3.2 Example 2: Performance planning
As a storage administrator, planning for new workloads and correcting bottlenecks are frequent tasks. This next
example will demonstrate how a solution may be designed to adequately plan and correct bottlenecks. In this
situation, the first example of the over utilized pool is used to determine the best solution for fixing the performance
issues. First, the Experimental Analysis panel is used to indicate how many more IOPS are needed.
3.2.1.1 Calculate the IOPS shortage
Using the previous examples Experimental Analysis display before the fix, notice the estimated maximum IOPs for the
default pool.
Figure 25 Estimated maximum IOPS for the PS6210E with RAID 6
Also, notice the total average IOPS that represents all of the I/O arriving at the PS Series group.
Figure 26 All IOPS for the PS Series group
To estimate the additional IOPS needed, subtract the total average IOPS (3349) from the estimated maximum IOPS
(991) to find the difference (-2358). So our goal is to accommodate the shortage of 2358 IOPs. These IOPs are
considered the total front-end IOPS needed.
Additional front-end IOPS needed = Total average IOPS – Estimated maximum IOPS
The delta we need to support is at least 2358 IOPS coming from the hosts at the same average, 70% reads and 30%
writes, with an 8KB I/O size. I/O size is important since 8KB is the typical size that the rule of thumb methods use for
sizing. Next, determine the back-end IOPS (total disks IOPS).
The formula below shows how back-end IOPS are approximated.
Total disk IOPS = reads + (writes x write penalty)