timothy.tatum (3/15/2016)
Hello,I am looking into automating my cube builds using SSIS and/or scripts and have a few questions.
1. I have a cube with yearly partitions and a Date Dim that has attribute indicators by day for "last 30 days", "last 60 days", etc. Obviously as the current date changes, the indicators at the day level will change as well. I realize I will have to "process full" the Date Dim daily, but will I have to process a previous year partition whose date range falls into the "last 60 days"? And then keep on processing that partition until "day 61" of the current year?
Yes, otherwise you will end up with the same facts in both partitions because they are not effectively moving.
2. I have a Postal Codes Dim with over 2 million entries that I intend to update only once a year. After I load the new postal codes and fully process the cube, can I remove the Postal Code Dim from the dimension update step once I start to process daily again (i.e., processing only the last partition)?
Yes.
3. I have a cube with 4 measure groups, but yearly partitions on just one measure group as the 3 others do not have nearly the number of records. If I process the last yearly measure group in a script, do I then need to "process full" on the 3 other measure groups or on the 3 other single partitions in those associated measure groups? I plan to use the Analysis Services Processing Task in SSIS for that step, but not sure whether to include partitions or measure groups.
Processing a measure group will automatically process all the partitions within, so you can choose to do it at that level if you only have one partition.