Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Plain View
Hive, mail # user - Bucketing float and double affects join?


Copy link to this message
-
Bucketing float and double affects join?
Nguyễn Thi Dương 2012-12-30, 06:53
Hi,

I understand that we can join float and double and they will be
automatically cast to the same types. However I want to use bucketing to
speed up the joins. If one of the joined columns is float and the other is
double, and the number of buckets are the same, will I be able to use the
hint "mapjoin" to speed up the join?

For example, my query is:

insert overwrite table logs_2 select /*+ mapjoin(u) */ l.*, u.highwayid
from logs_bucketed l join unique_longlats_mapped_bucketed u on l.lon u.lon and l.lat = u.lat;

For example here, l.lon is double, and u.lon is float.

When I ran this query, no rows were inserted into logs_2, so I guess
there's something wrong here.

Thanks