Wednesday, June 2, 2010

Does PowerPivot compress smaller files efficiently?

The Analysis Services engine shipped with SQL Server 2008 R2compresses and processes the data which is loaded by the Analysis Services into the worksheet. All this happens in the background using the Power Pivot [Microsoft.AnalysisServices.Modeler.FieldList.Addin.Integration] menu option. The storage mode of this service called the VertiPaq  does the magic of manipulating columnar data in memory.

Results from a simple test of compression:

I downloaded the Power Pivot sample file in Excel format from the CodePlex site. Review my recent post here for details:

  • The file is called ContosoStoreData with columns from A to S with 307 Rows as shown here. The file size is 88KB on disc.

  • I copied the data and pasted into an excel spread sheet (not using the PowerPivot)and saved it as ContosoNormal. The file size of was 48KB on disc

  • Next I imported ContosoNormal into an Excel Spreadsheet under PowerPivot control and named the new file, ContosoNormalReverted. It had the same number of columns as ContosoNormal and the same number of rows(307). The file size was 52KB on disc.
My computer was an Acer Notebook with the following specifications: Windows 7 Ultimate 32bit OS with 3.00 GB RAM. U2700@1.3 GHZ. Free space of 33 GB available on the 288 GB hard disc.