Relating production workloads to LSPR workloads

However, as discussed in the LSPR Workload Categories section, the underlying performance sensitive factor is how a workload interacts with the processor hardware. These past techniques were simply trying to approximate the hardware characteristics that were not available through software performance reporting tools. Beginning with the z10 processor, the hardware characteristics can now be measured using CPU MF (SMF 113) COUNTERS data. Thus, the opportunity exists to be able to match a production workload to an LSPR workload category via these hardware characteristics (see the LSPR Workload Categories section for a discussion about RNI – Relative Nest Intensity).

The AVERAGE RNI LSPR workload is intended to match the majority of customer workloads. When no other data is available, it should be used for a capacity analysis.

DASD IO rate has been used for many years to separate workloads into two categories: those whose DASD IO per MSU (adjusted) is <30 (or DASD IO per PCI <5) and those higher than these values. The majority of production workloads fell into the “low IO” category and a LoIO-mix workload was used to represent them. Using the same IO test, these workloads would now use the AVERAGE RNI LSPR workload. Workloads with higher IO rates may use the HIGH RNI workload or the AVG-HIGH RNI workload that is included with zPCR.

For z10 and newer processors, the CPU MF data may be used to provide a more accurate workload selection. When available, this data allows the RNI for a production workload to be calculated. Using the RNI and another value from CPU MF, the L1 cache misses per 100 instructions, a workload may be classified as LOW, AVERAGE or HIGH RNI. This classification and resulting workload selection is automated in the zPCR tool. It is highly recommended to use zPCR for capacity sizing. For those wanting to perform the workload selection by hand, the following table may be used for z10, z196, zEC12, z13, z14, z15, and IBM z16 (note L1MP stands for L1 misses per 100 instructions and is a value that may be calculated using the CPU MF counters data):

Note this table may change in the future.