z/VM system load during the relocation process

This section shows the typical z/VM CPU usage during live guest relocation.

Figure 1 shows the typical CPU usage by z/VM during the live guest relocation. In all cases shown in Figure 1, an optimal number of FCTC connections per FICON channel are configured. The number of FICON channels is varied from one to seven.

These values were obtained using the “Processor Activity” log (PROCLOG, FCX144) of the z/VM Performance Toolkit. In Figure 1, the values for CPU 0 are shown; the other CPUs exhibit similar usage patterns, but with slightly less values.

Figure 1. z/VM usage of CPU 0 during live guest relocation for various ISFC logical link configurations
Graph of z/VM usage of CPU 0 during live guest relocation for various ISFC logical link configurations

Observation

The peak CPU usage afforded by z/VM for performing the live guest relocation increases with the amount of FICON channels involved; however, at the same time the relocation times decrease.

Conclusion

The area below each graph is an indicator for the overall CPU consumption spent by z/VM for the relocation process. For the configuration depicted in Figure 1, the largest area results when one FICON channel is used, and a relative minimum is assumed when four FICON channels are used. This might be interpreted as a hint that – in the context of the given workload and the general setup of the test environment – the four FICON channel configuration provides for the most economical approach with respect to the z/VM CPU consumption for the relocation process.