System BufferPool Data.....2

GetPage Total 1.061M

Seq 848511

Random 212508

SyncRead Seq 60914

Random 179327

SeqPrefRead 25309

Request 26123

PagesRead 775415

Dis NoBf 0

Dis NoEn 0

WKFAbort 0

LstPrefRead 286

Request 249

PagesRead 3739

DynPrefRead 607

Request 626

PagesRead 18815

DWHorThreshold 1

DWVerThreshold 19

DM Threshold 0

SynHPool Read 0

Write 0

AsynHPool Read 0

Write 0

DAMoverReadS 0

ReadF 0

WriteS 0

WriteF 0

Same Pool, BP0

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

Notes:

This slide contains the rest of the information for BP0; although it has been wrapped into two columns to reduce the number of slides.

Take special note of the ‘Synch Reads Seq’ field. First, there will almost always be some counts in here because prefetch always issues 1 synch I/O when it starts to run. However, large counts here, especially when coupled with low or ‘negative’ hit ratios indicate a serious performance problem. Pages read into the pool for prefetch can be released (or thrown out) by other prefetch activity before an application can get to them for processing. When this happens, they are read back in using synchronous I/Os...and increment this counter. This has a serious performance impact on the elapsed time for the requesting, and most other, applications. This is most likely to happen in three (but not only) situations: when the processor is too busy (> 95%), when the application MVS dispatching priority is too low, or when the processor is not very busy and several large scan jobs execute concurrently.