Перейти к содержанию

Заметки

That Old Restart Problem Strikes Back: Getting What You Pay for

During my previous research, I was also testing the same UPDATE statement against a partitioned table. I was slightly puzzled that I did not observe any statement restarts in the performance statistics, even after I updated tens of millions of rows in the partitioned table. It turns out that there are still statement restarts there, however, they do not lead to such dreadful ramifications when a table is being read three times.

That Old Restart Problem Strikes Back: Setting the Stage

When I was reading about statement restarts, I thought it usually happens in a multi-user system or in the presence of triggers. However, last week one of our developers asked me to help him to figure out why a simple update statement was reading far more data than it was in the table. That blog post is published to share my findings about that.