Hi,
"a complete RP support ..."
RP is not fully supported on DB2, there is no range stats usable by the DB2 optimizer, therefore access plans is based on "Nb rows / Nb partitions used". This gives inaccurate estimations.
P1 = 1 million
P2 = 1 million
P3 = 0.1 million
select * from tab where part = P3 ==> card estimates: 2,1/3 = 0.7 million.
Oracle estimation: 0,1 million
RP is not supported on DB2 BLU. Compression column dictionary is based on 5 million rows. With a 13 billion row table, the "manual split" using UNION ALL views show that storage savings is 18 % ! a dictionary on smaller tables is better.
When will we get RP full support ? This feature seems to me more important that having "automatically created column aliases" on CTE tables using functions.
Thanks,
------------------------------
Jean-Marc BLAISE
------------------------------
#Db2