Hi Igor,
afaik in the manuals there's not much details on the internal implementation of OLAP functions.
- OLAP functions might skew in spool, cause the data is distributed based on the PARTITION/ORDER columns.
- Another downside is the spool usage, when you Explain it you'll see two spools, i.e. you need twice the spool size when you add an OLAP function.
But in most cases OLAP functions other than the GROUP WINDOW stuff can't be rewritten using old-style syntax (of course rewriting is possible, but the SQL is *much* more complicated and performance is horrible), so you can only try to spot if the usage of an OLAP function is not really needed in this specific case and can be replaced by a simpler construct.
Dieter
↧