I have read about the benefits of turning on the registry variable DB2_USE_ALTERNATE_PAGE_CLEANING which became available in v8.1.4. What I cannot find is any advice on how to measure the performance of this registry variable during testing.

One thing I can think of is that the peak I/O wait time might be reduced since the I/O cleaners will be performing I/O more evenly instead of waiting until a threshold is exceeded, causing heavy bufferpool cleaning to be triggered.

Besides looking at data from the iostat command for before/after comparison, is there a good way to measure and compare metrics to determine exactly what benefit is being received by turning on this registry variable?