Tip of the week: Enhance performance on mapping, group by and where
Working for a customer with large qvd-tables, 5-15 million rows and "wide" tables (many fields), I have found these performance improvements recently.
If you do a mapping load from a large qvd-table this is slow.
Instead if you load the two mapping fields into a temporary table and then do this mapping load from the temporary table you will see great performance improvements. First the load into the temporary table is optimized and secondly the load from a small (few fields many rows) resident table is way faster than loading from the qvd-file. In my case 10 minutes to under 10 seconds per mapping table.
I have found the same strategy to be working fine if you need to use a "where" and "group by" statement and only want a few fields from the lagre qvd-table.