Step 1. - Loading a 1 single day with 5 million records, 4 dimensions, 20 expressions with a time of 1.5 hours and QVD generates a 800 MB (per day).
Step 2. - Read the last 2 months of qvds generated in Step 1 and the size of this qvw is 22 GB of disk space, and processing time as 32 GB of RAM
on the server, with a time of 4 hours. This is time and space, so try making a "GROUP BY" in Step 1 but make the mistake,
"OUT OF VIRTUAL AND / OR LOGICAL MEMORY"
Do you have any idea to reduce the size of QVD, and not out of memory, with somewhat similar to "GROUP BY"?
This is an example of GROUP BY to try running.
SUM(Amount) AS Amount
DROP TABLE TEMP