Re: Index creation takes more time?

"Herouth Maoz"  writes:
> So, I must, at this point, draw the conclusion that string comparison is a much, much heavier task in utf-8 than it is in an 8-bit encoding - or that the collation is the problem.

Going from "C" collation to anything else is generally a huge hit in
terms of string comparison/sorting performance.  Do you really need
locale-aware sorting, or just enforcement of utf8 encoding?  Because you
can use "C" locale with any encoding.

			regards, tom lane