Re: 4.1 ten times slower when using utf8 charset

[Harrison Fisk]

Hi,

latin1 is almost always faster than UTF8 in general as you have noticed. The reason for that is because UTF8 has much more complex collations and rules on how comparisons occur. The speed difference depends mostly on what collation you have set. What collation did you use with UTF8?

There are some other optimizations you can do as well, but without knowing your schema or queries, I can't help you at all with those.

Now as to using latin1 instead of UTF8 for japanese, keep in mind that comparisons and sorting won't neccesarily work correctly. So it might be faster, just not correct all of the time. The making it correct is what makes it slower.

Harrison Fisk, Trainer and Consultant
MySQL AB, www.mysql.com