Ive been working with cognos from a year now but I stumble into another problem that needs urgent workaround resolution. I have this model that automatically generates cube (using a batch file) every day. the generation is successful although the main issue is the time it consumes to generate the cube. Although it is somehow acceptable considering the data involved (data includes historical transactions from about 7 years ago). Now the main idea we came up is to limit or somehow include the data from previous years in the model and be remained untouched in the everyday cube generation (since there are no expected changed int these data). Only data from current year should be processed and these will be added with the historical data. So far I have tried using manipulating iqd but the results do not include the historical data in the generated cube. Also I am using both 7.4 and later version ibm cognos 10 so I wish you could give me ideas on how to work on with this on both versions.
Asked
Active
Viewed 1,127 times
1 Answers
0
What you're looking for is time based parititioning. Search on transformer time based partition in Google. For example this is for 8.4 but you should be able to find doco on v10.
This allows you to build your prior years once, which will exist as multiple individual MDC files. You then only regularly build your current year cube, which refreshes just one MDC file. These are all transparently sown together in a master dummy MDC and VCD file.
Apparently v10 has some build optimisations so I would do this in v10 rather than v7 if possible
Nick.Mc
- 18,304
- 6
- 61
- 91