This would do a fast solve of just the data for the new day, which would get added to the previous data in the cube.
BEGIN DBMS_CUBE.Build ( 'DATA_OLAP.DAY_BAGS_NFD_CUBE USING (LOAD NO SYNCH, SOLVE)', 'S', -- refresh method - Fast Solve FALSE, -- refresh after errors 0, -- parallelism. NB - This may need to be changed (or at least noted down) during testing TRUE, -- atomic refresh TRUE, -- automatic order FALSE); END;
1) Create a materialized view log on the base table and specify 'F' as the refresh method in DBMS_CUBE.BUILD (or DBMS_MVIEW.REFRESH).I already have an mview log on the base table
(2) Partition both your cube and your fact table by DAY and specify 'P' as the refresh method.Both are already partiioned by day
(3) Specify a refresh method of 'S'. This will load the entire fact table into the cube each day, but only the latest partition will be re-aggregated. This is called the 'Fast Solve' method.I thought that this was what I was trying to achieve, but using the data dimension to get it only to load and see the current day. The Cube MVs causing a problem as initially specified.