Query I/O Optimization


Reading of multiple Items' data from Store whose data is stored in separate data segment can be further optimized by doing delayed bulked data reads.

By submitting multiple Items' data segments for reading to the Store Reader, the Reader can identify which segments are contiguous, create a bigger segment for reading asynchronously and thus, reduce the "small block" reads which is not performant. Typically, SOP Items' data values in data segment are co-located, thus, this optimization will increase query I/O to optimal level.

No data structure change required, 'just ability to take advantage of the Query I/O pipeline.
Closed Nov 14, 2015 at 7:17 PM by grecinto
This issue was just used to document the feature currently in place. Closing...