Recently I came across one of suggestions by Microsoft Advisory consultant regarding optimization data write-back pattern for the SSAS cubes. It was recommended issuing CREATE CACHE command following by UPDATE CUBE for the same cell/subcube definition:
CREATE CACHE FOR [MYCUBE] AS ("Cross join of all the dimension members defining subcube to be updated")
UPDATE CUBE [MYCUBE] SET ("Cell") = 100, ("Another Cell") = 200,... etc. for every cell.
This seems quite weird, but based on our benchmark the entire process (CREATE CACHE and UPDATE CUBE) takes less time than just UPDATE CACHE.
Folks, can someone advice on the insights of the process? Shell this approach to be suggested for an enterprise-level planning application?
View Complete Post