The following post was targeted to be an index holder for the document involving Review of Application Performance for BCU (TP 12886). I will be adding extra chapters as we find new performance improvement methods.
Log of performance improvement methodologies
We had good amount of discussions with IBM and within our group and slowly added new techniques to a few netting procedures. The idea was to remove the cursor build code that was known to eat up lot of memory in both holding the data and for sorting them (when order by clause was used).
1. We used the ‘exists’ predicate in matching one on one for a set of data, I have copied the IBM information center’s comment on the syntax and how it works.
2. Global temp tables were used (very similar to cursor, but we got great performance improvements by coupling this with other advantages the ‘table’ in DB2 gives, compared to cursors.)
3. Table partitioning (range or hash)
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment