![]() | ||||||||||||||||||||||||
Deploying SSIS projects on a large database can be simple - unless a sort is needed. A project which includes a sort often hits performance limits once the data grows beyond a gigabyte or two. But it doesn't have to be that way. NSort brings world-record sorting technology to SSIS, cutting sorting time by 70% or more while eliminating the need to restructure the project into RAM-sized partitions. By carefully using temporary files NSort easily processes data sets dozens or hundreds of times the size of memory. Users typically find that data is processed as fast as it is read, while using even less cpu time than the standard sort. The following table shows the elapsed time necessary to sort various data set sizes using NSort and the standard sort component. In all cases the data consists of two ascii columns, 10 characters and 90 characters in length. The first column is used as the sort key in the US English locale. The unsorted data is read from a file using the Flat File Source component. The sorted data is written to a file using Flat File Destination.
NSort for SSIS easily substitutes for the standard Sort component, not requiring any other changes to your projects. Projects can be implemented naturally, simplifying the logic of the data flow.
NSort can accept multiple input streams, and can partition its output data into multiple output streams that are emitted in parallel, facilitating parallel downstream processing:
The following is a common customer reaction when using NSort for SSIS:
For more information, see the NSort for SSIS Manual or, better yet... Both x86 and x64 versions are currently available for SSIS 2005 and SSIS 2008. NSort for SSIS is a separate product from the Nsort executable and library, and is priced separately. |