Can you import data without rebuilding the cube?

[manlypeeps]

We are considering implementing MS Analysis Services but were wondering if it is necessary to rebuild the cube each time you import data; eg if a retailer wanted to append a day's trading data to an existing cube, would they have to rebuild the cube each time?

Thanks very much,

Steve

 

[MDXer]

Yes. But depending on the cube design and the volume of data the cube process could take a few minutes. You need not build it completely but rather reprocess the cube or dimension, as long as the source is the same. your fact data can have multiple source tables using partions.

"Shoot Me! Shoot Me NOW!!!"
- Daffy Duck

 

[johnherman]

Incremental updating of cubes is supported, and works very well in association with partitions.

Sometimes the grass is greener on the other side because there is more manure there - original.

 

[MDXer]

A heads up on incremental updates. Incremental updates are very useful in dealing with reprocessing a cube but care should be taken. Since only new data is built then merged into the existing cube the indexes within the cube can and do get out of order having an affect on preformance of queries. The performance impact may be unoticable in many cubes but with large complex cubes it can become more noticable. Giving this an occasional Full reprocess of the Cube partitions probably should be done. The approach I have taken is incremenntal processing until a new partition is started once data is no longer being processed into a partition I then do a full process of the dimension and then leave t alone.

"Shoot Me! Shoot Me NOW!!!"
- Daffy Duck