Performance of Sort-Based Operators in View Models

Several operators must sort their inputs. These include Order By, Join (only when using the Sort Merge algorithm), Group By, and Intersection, as well as any operator that uses the Distinct option to eliminate duplicate rows.

WorkSpace Data Federation uses an O(n log n) algorithm to perform all its sorting operations. This algorithm writes temporary data to disk if the result set to be sorted is too large to fit into memory; thus it scales to accommodate arbitrarily large result sets.

When operators in your model sort their data, WorkSpace remembers which columns the data was sorted by and passes that information along to downstream operators in the view model. Consequently, operators that need sorted data don’t have to re-sort data unnecessarily. This feature can dramatically increase the performance of WorkSpace-generated data services.

Two factors that can affect sort performance are sort chunk size and disk space for temporary sort files.

• Controlling Sort Chunk Size
  To enable sort-based operators in data services to handle arbitrarily large result sets, Sybase Data Federation breaks the inputs into chunks and uses the local disk as a temporary backing store to hold intermediate results. You can control the size of the chunks in the view model.
• Providing Enough Space for Temporary Sort Files
  When a data service sorts data, it writes sort files, which can be quite large, to a temp directory. If the temp directory doesn’t have enough disk space, the data service fails. You can set the location of the temp directory.

For product-related issues, contact Sybase Technical Support at 1-800-8SYBASE. Send your feedback on this help topic directly to Sybase Technical Publications: pubs@sybase.com