Creating Analysis Services Cubes - SQL Server Analysis Services 2012 Cube Development - page 122

each partition limits the rows by OrderDateKey . A more effective way of partitioning
this measure group would be to specify the range of SalesOrderNumbers
( SalesOrderNumber is the DISTINCT COUNT column) to be included in each par-
tition. The following queries could be used to define four partitions based on the
SalesOrderNumber column:
SELECT * FROM FactInternetSales
WHERE
SalesOrderNumberBETWEEN'SO43697'AND'SO55999'
SELECT * FROM FactInternetSales
WHERE SalesOrderNumber
BETWEEN'SO57000'AND'SO63999'
SELECT * FROM FactInternetSales
WHERE SalesOrderNumber
BETWEEN'SO64000'AND'SO69999'
SELECT * FROMFactInternetSales
WHERE SalesOrderNumber>'SO70000'
If we expect the majority of our queries to check the number of orders by a calendar
year, a more effective strategy would be to partition the measure group by the sales
order number as well as the order date.