I have the following table containing 2.000.000.000 rows:
CREATE TABLE [VeryLargeTable] ( [dataCol1] [NVARCHAR(MAX)] , [dataCol2] [NVARCHAR(MAX)] , [Manufacturer] [NVARCHAR(10)] , [MsaiSalesType] [NVARCHAR(2)] , [Return Volume] [NVARCHAR(2)] )
The data is distributed this way:
I have to determine the first column(Sales Types), which is determined by the following 3(Manufacturer, MsaiSalesType, Return Volume).
The table will be created on the spot, then the data will be inserted via SSIS from a flat file. After that, data will not be inserted, updated, deleted.
I have taken into consideration filtered indexes and partitioned tables.
Would they be efficient if the data is so un-evenly distributed ? What are my other options ?
If I get rid of the data representing 95 % of the table, I should have no further trouble filtering the rest of the data.