Use of processing weights
As an example, consider a system with 6 CPs and 3 LPs defined as follows:
LP Name Logical Cores Weight
------- ----------- ------
ZVSE 1 300
ZOSTEST 6 100
ZVM 2 900
Processing weights can range from 1 to 999 (weights of less than 2% difference are not considered significant) and are used as follows:
- The processing weights for all active, sharing LPs are added together. This total is considered to be 100% of the processing resource available to shared CPs.
- The share of processing resources for each LP is calculated by
dividing the processing weight for each sharing LP by the total processing
weight. For example, at peak CP utilization levels, the dispatcher
allocates shared processing resources to each of the LPs as follows:
ZVSE 300/1300 = 23.1% ZOSTEST 100/1300 = 7.7% ZVM 900/1300 = 69.2% - The share of processing resource for each online logical core
with HiperDispatch disabled in the logical partition is calculated
by dividing the share for each LP by the number of online logical
cores. The share for each logical core is as follows:
ZVSE 23.1/1 CP = 23.1% ZOSTEST 7.7/6 CPs = 1.3% ZVM 69.2/2 CPs = 34.6%These percentages are used to determine preemption priority for I/O interruptions. A lower priority logical core can be preempted when an I/O interruption is pending for a higher priority logical core when the following occurs:
- The higher priority logical core is further behind in its share, or
- The higher priority logical core is not as far ahead of its share as the lower priority logical core.
For example, the lower priority LP is receiving 15% more than its processing share, and the higher priority LP is receiving 10% more than its processing share.
As long as there is excess CP capacity, processing weights have no effect on the CP resources consumed. Weights affect processing when the number of logical cores that need processing time is greater than the number of physical cores available.