Hi,
One of our customer is using CHAR and collation=fr_FR@colstrength=3 , and find that performances are not as good as when they use VARCHAR and the default collation.
I can't really see the difference , but is there any reason why there would be one ?
And if there's one , can we measure it ?
Regards
Bernard
Collation and performances
+2
Vertica default locale is en_US@collation=binary, which uses binary collation.
Binary collation is faster as V compares binary representations of strings.
When locale collation is non-binary, the GROUP BY on string data will make V to call COLLATION even if the function is not specified in the query.
The reason for performance degradation is the sorting time, because V sorting depends on LOCALE.
+2Many thanks Moshe
Have you got any idea of the overhead ? I mean in term of magnitude : is it twice slower or just adds a few microsecond to each query ?
Depends on the use-case:
The amount of rows, predicate columns number and VARCHAR(s) variable size.
Sign up
Already have an account? Login
Welcome to the Rocket Forum!
Please log in or register:
Employee Login | Registration Member Login | RegistrationEnter your E-mail address. We'll send you an e-mail with instructions to reset your password.
77 4th Avenue
Waltham, MA 02451 USA
