IYO, which is better, putting conditions on queries to limit the data returned to the cube, or not using conditions and filtering all the data returned to just show the data you need? This question really only applies to large 500K plus chunks of data.
I’ve read many threads on this forum about conditions on non-indexed columns and also poor performance in Reporter with too much data in the cube but would value your opinions on which is the preferred method.
I always try to bring back only the data required.
I’ve got 512MB RAM on a P4 2.8, but other users may only have 128MB on a P4 1.5, so the end user PC spec should be one consideration.
That and recalculation and filtering speed when dealing with large amounts of data when you initially create the report.
As a rule of thumb I prefer conditions in the query panel (limiting the initial result set). Poor performance can usually be improved by database optimizing such as adding new indexes etc.
By the way, I would prefer a more descriptive title for this topic such as “Query conditions vs filters” instead of “Which is better”
I guess it is not always possible to put indexes on all the columns you want to use as conditions. It is not logical to put indexes on textually descriptive columns such as Surname or Customer as these then have an impact on the ETL processes that get the data there in the first place and also on capacity as they create pretty large indexes!!
(BOB member since 2004-07-05)
(BOB member since 2002-06-20)