In my last two posts, I discussed two scenarios - one involving updates and another involving large objects - where SQL Server extends the duration of read committed locks until the end of a statement instead of releasing the locks as soon as each row
