• GilaMonster (5/9/2013)


    Correct, it's CPU and RAM, but how does killing a query that's done a lot of physical IO help? The IO's already done by the time you kill the query, the stuff that it read already in memory.

    Those ad-hoc queries are hammering the SAN really bad (tempdb and the actual Data LUN) and they keep running for ever, affecting other clients as well.

    Killing those ad-hoc queries after X minutes will release the IO pressure. So instead of hammering the SAN for 5+ minutes, if they are not done my 1 min or less , they will be killed.