• Eric M Russell (8/7/2014)


    Even if the QA environment only has 10% the data volume as production; so long as the object schemas, configuration, and data cardinality are equivalent, the execution plans should be equivalent too. So, for example, if adding a new join and columns to a stored procedure results in 30% more page reads in QA, then you should expect the same proportional 30% increase in production. That's what you should focus on, how the latest changes modify the the plan and proportionally impact i/o, cpu, and memory, not runtime duration.

    Try offering to management the answer that runtime duration isn't important when end users are seeing slow responses after a recent change. It won't go down well!