Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
tybaltzeke
Starting Member
28 Posts |
 Posted - 2009-10-19 : 21:00:09
|
| Hello...I would like to hear opinions on an issue I am facing with a client. --They have a warehouse with millions of records. They are hoping to be able to refresh this incrementally through ETL on a daily basis although they are no where near that objective currently.-- They are building a cube that they also want to process intra day, to move toward a realtime environment. In order to speed up the processing time they have used a sparse design for the cube with almost no aggregations and heirarchies with 'missing levels', for example the time dimension has year and month but not quarter, so quarters have to be calculated in MDX.Here are my questions What would be the real advantage of re-processing a cube on a warehouse that is out of date. How much would the report generation performance be affected by the lack of aggregations in the cube. If they put in aggregations, is it possible to incrementally load the cube's aggregations so that the aggregation calculations reflect the incremantal processing.. If this is possible could some one direct me to a source on how to do this. For example, if you use ROLAP for today's data and MOLAP for older data, does the entire cube have to be reprocessed anyway so that the aggregated calculations are correct? Thanks for any help you can provide. |
|
|
|
|
|
realtime processing and query performance q's