Here you find information related to savepoint handling. The file is useful to specify the amount of data that has to be handled for a savepoint, write times are supplied and thus you can retrace the duration of savepoints. Additionally detailed information about the different savepoint phases is delivered.
Content - Description of the collected values
|CacheFlushWrites||physical writes for cache flush|
|CacheFlushWrittenPg||Number of pages written for cache flush|
|AvgWTime_CacheFlush||Average.time (ms) of cache flushes|
|Savepoints||Number of savepoints||Default metric|
|SvpRuntime||runtime of savepoint(s) in s||Default metric|
|SvpIOCalls||Number of IO calls|
|SvpWrittenPg||Number of pages written|
|SvpQueueingForIOTime||runtime of queueing for I/O in s|
|SvpClusterSortCheckTime||runtime of cluster sort checks in s|
|SvpPhase1Flushtime||flush time of phase 1 in s|
|SvpPhase1IOCalls||Number of IO calls in phase 1|
|SvpPhase1WrittenPg||Number of pages written in phase 1|
|SvpPhase2Flushtime||flush time of phase 2 in s|
|SvpPhase2IOCalls||Number of IO calls in phase 2|
|SvpPhase2WrittenPg||Number of pages written in phase 2|
|SvpPhasePrepareTime||datacache locked time in s|
|SvpPhase3Flushtime||flush time of phase 3 in s|
|SvpPhase3IOCalls||Number of IO calls in phase 3|
|SvpPhase3WrittenPg||Number of pages written in phase 3|
Examples of Analyses
- <Links to BCP tickets>
- <Links to new pages or WORD documents describing an analysis>
Helpful with ...
|869267 - FAQ: SAP MaxDB Log area||Provides information about the use of savepoints.|
|819641 - FAQ: SAP MaxDB Performance||Savepoint duration is related to disk I/O.|
|936058 - FAQ: SAP MaxDB Runtime Environment||Find some information about which task types participate in savepoint handling.|
Expert Session (Link to Trainings)
Content / Helpful with ...
|Session 10: SAP MaxDB Logging||Describes the concept of transaction handling with savepoints and how to influence and trigger them.|
|Session 26: SAP MaxDB I/O Concept||In case of long running savepoints I/O might be the bottleneck.|
Find more about I/O mechanisms.