The RMFCRL query produces charts that show cryptographic coprocessor use by workloads executing on z/OS LPARs running on IBM z10 and later generation Central Processing Complexes (CPCs), such as zEC12. In addition to showing cryptographic function use, these charts quantify delays incurred because coprocessors were busy, and the average number of wasted machine cycles lost due to these delays. The RMFCRL query creates the following CSV output file:
Crypto Coprocessor LPAR Level Activity
The data extract generates the following three chart views:
Chart view #1 displays overall cryptographic function use by type:
Advanced Encryption Standard
Data Encryption Algorithm
Pseudo-Random Number Generator
Secure Hash Algorithm
The overall count of each function type executed is displayed using stacked vertical bars and the left Y axis. The average LPAR Individual Processor Unit (IPU) busy is displayed using a horizontal line and the right Y axis. The first chart sample above was produced by inputting two cycles (days) of the DETAIL timespan HARCML file. This chart, for LPAR ‘SY36’ running on an IBM 2097-729 (z10 CPC with 29 CP engines), shows that for each of the two days, cryptographic coprocessor function execution starts ramping up about 7 AM, peaks between 11 AM and 4 PM, then decreases through the late afternoon and evening hours. Two cryptographic functions are executed by the workloads running on LPAR ‘SY36’, DEA and SHA, with the vast majority being SHA (green bar) functions. The average logical IPU busy percentage, which represents the percentage of available machine cycles used by dispatched logical processors, never exceeds 12 percent.
Chart view #2 displays the percentage of each function type that was blocked from execution, using stacked vertical bars and the left Y axis. Crypto functions blocking occurs when the cryptographic coprocessor is busy. The second chart sample above, shows that for each of the two days, the blocked function percentage was very low, far less than 1/10th of 1 percent. Even though the first chart showed that the majority of crypto functions executed were SHA functions, the majority of blocked functions were DEA
Chart view #3 displays the average number of penalty cycles incurred per function block, for each function type using vertical (non-stacked) bars and the left Y axis. Penalty cycles are wasted machine cycles expended while the function is waiting for the cryptographic coprocessor.
By default, the query is executed against the DETAIL timespan HARCML file, and each X axis value represents data from an RMF interval. The X axis metric, DATE HOUR MINUTE, is constructed by concatenating the DATE, HOUR, and MINUTE from the end time boundary of each interval.
|
Copyright © 2014 CA.
All rights reserved.
|
|