May 2019
Intermediate to advanced
484 pages
11h 23m
English
Optimizing I/O is usually the best tuning approach, because with higher I/O comes higher memory consumption as SQL Server needs to store more data pages in the buffer pool, and higher CPU as cycles are spent processing I/O requests and data movement.
Actual I/O Statistics* provides information on Large Object (LOB), physical and logical reads allowing for immediate insight into the cost of an operator, without the need to collect or interpret the information from SET STATISTICS IO.
If the query was executed in parallel, we can see how many data pages were read by each thread. In the following screenshot, we see the detail for the most expensive operator in the preceding plan, a clustered index scan:
Read now
Unlock full access