Disable Logging during build to improve overall load
performance. Once load is complete turn it on.
If you choose to set NOLOGGING for the LOB segment (of
the AW$ table containing your Analytic Workspace) check out
MOSC 1058851.6 for information pertaining to event 10359
which can reduce I/O for frequently updated NOLOGGING LOBs.
* Turn Logging off prior to commencing build.
GLOBAL.AW$GLOBAL MODIFY LOB (awlob) (CACHE READS NOLOGGING);
* Turn Logging on after build completes.
GLOBAL.AW$GLOBAL MODIFY LOB (awlob) (CACHE READS LOGGING);
SQL INSERT DIRECT is up to 53x faster than SQL INSERT.
NOTE: Source tables should be ordered by the inverse
of the target object's dimensionality when loading data from
relational sources into an analytic workspace.
SQL IMPORT is up to 7x faster than SQL FETCH.
Before using or recommending these parameters you
should fully understand how they work. The defaults should
work fine in nearly all cases.
Size of the olap page pool in bytes. Leaving it unset
or setting it to 0 would turn on auto dynamic page pool
Rate of pool decrease as a percentage of current pool
size. The default is 50%.
Rate of pool increase as a percentage of current pool
size. The default is 20%.
Pool size high watermark, a percentage of PGA
Aggregate Target. The default is 50%. It won't grow beyond
Pool size low watermark in bytes. It won't shrink
below that point. The default is 256k.
The point at which OLAP will shrink the pool, a
percentage of PGA Aggregate Target. The default is 90%.
Page cache hit ratio target. Default is 100%. (Not
much need to change.)
NOTE: These hidden parameters don't apply to MTS mode.
To view these parameters run the following select
statement as SYS:
col par format a28 heading 'Parameter' col val format
a8 heading 'Value' col des format a36 heading 'Description'
select ppi.ksppinm par, psv.ksppstvl val, ksppdesc des from
sys.x$ksppi ppi, sys.x$ksppsv psv where ppi.indx = psv.indx
and ppi.ksppinm like '%olap_page_pool%' order by ppi.ksppinm;"