Thank you very much!
Zhigang -----Original Message----- From: Morgan Tocker [mailto:morgan.toc...@oracle.com] Sent: Wednesday, February 12, 2014 10:16 AM To: Zhigang Zhang Cc: mysql@lists.mysql.com Subject: Re: LIKE sql optimization Hi Zhigang, On Feb 11, 2014, at 8:48 PM, Zhigang Zhang <zzgang2...@gmail.com> wrote: > I want to know the reason, in my opinion, to scan the smaller index data has > better performance than to scan the whole table data. I think I understand the question - you are asking why MySQL will not index scan, find matching records, and then look them up rather than table scan? I believe the answer is that there is no way of knowing if 1 row matches, or all rows match. In the worst case (where all rows match), it is much more expensive to traverse between index and data rows for-each-record. So a table scan is a “safe choice" / has less variance.=