Object Statistical Analysis

Statistics for Table Space.......................DB2PO.SOLNSORD

Number of GetP...............690,589 29.0% of pool GetP

Number of Sequential Access.634,019 91.8% of GetP

Number of Random Access......56,570 8.2% of GetP

Number of RID_List................0 0.0% of GetP

Number of Random Misses......27,438 7.6 Misses per Sec

Number of Misses (others)....19,740 5.5 Misses per Sec

Number of No_Reads................0 0.0% of GetP

Number of Hits..............643,411 95.2% of GetP (Appl. HIT RATIO)

System HIT RATIO...............-1.9 %

Avg. Page Residency...............0 Seconds

Previous slide Next slide Back to the first slide View Graphic Version

Notes:

After the overall pool statistics, we want the same type of information for each object within a pool. The objects should be sorted in order of decreasing Getpage activity, because we want to see the most heavily accessed objects first.

The V3 Buffer Pool displays can produce similar information, but can’t order it. Therefore, the most heavily accessed objects might be at the bottom of thousands of lines, or might be truncated off the report. Likewise, it will not work to try groups of display commands. At the detail level, they are incremental since the last display...and you need all information from the same time span and duration for the results to be meaningful.

Online monitors can collect the data, but cannot save and process enough of it for the types of meaningful statistical analysis and other processing we have in mind.

Snapshot collections of a ‘few minutes’ are not generally useful unless the transaction workload and mix is very consistent...over a short period. Longer periods provide much better consistency and accuracy. Typically, I like to get more than two million getpage records (198s) for a V3 system...since almost half of these are ‘release page’ records. In reality three million plus is a better collection and provides the best consistency.

Analysis continues on the next two slides.