Recommendations for projects

Each project needs to perform an evaluation of these tests and their results in order to determine whether or not modifying indexes is appropriate given the database size, performance requirements, and available disk space.

Changing indexes may have little or no benefit to projects that have small amounts of TimeSeries data, or have bad index fragmentation on their collect table. Testing has not been performed to determine where the performance improvement cutoff is.

Our tests resulted in a 22% increase in speed in reading from TimeSeries Collect using the readTimeSeriesCollect procedure, which may help projects that have issues with trending performance. Inserting new records was seen to be slightly slower, but the read performance may be the more important gain. Projects need to be aware that the size of the indexes may increase by approximately 30%, so the system needs to be sized appropriately before making this change.