This is partially a question of the SQL engine you are using, but I think all of them do a good job here.
So yes if you use the hash partition column in a predicate in your query and therefore Kudu scan, then only the tablets that match predicate will be scanned. On Wed, May 13, 2020 at 11:04 AM Ray Liu (rayliu) <[email protected]> wrote: > For example, I have a table with composite primary key (a,b,c) (in that > exact order) and column b is set as hash partitioning column. > > If I run a sql query like this > > select * from table where b = 1 and c = 2 > > > > In my understanding, because we already set b as hash partitioning column > Kudu will not do a full table scan. > > Instead Kudu will look for “c=2” in the hash bucket of “b=1”. > > Is my understanding correct? > > > -- Grant Henke Software Engineer | Cloudera [email protected] | twitter.com/gchenke | linkedin.com/in/granthenke
