公开(公告)号：CA2317081A1

公开(公告)日：2002-02-28

申请号：CA2317081

申请日：2000-08-28

Applicant: IBM CANADA

Inventor： HAAS PETER JAY ,  LOHMAN GUY MARING ,  RJAIBI WALID

IPC: G06F17/30

Abstract: Column cardinality for a table in a partitioned relational database is estimated. In each node in the database containing partitioned portions of the table a hashing function is carried out to generate a hash data set for each node, according to one of sample counting, linear counting or logarithmic counting techniques. The hash data sets are merged in a coordinator node to generate a combined hash data set. The combined hash data set is counted and correcte d according to the appropriate sample counting, linear counting or logarithmic counting techniques. The resulting count is adjusted for hash collisions and the resultant value is used to estimate the column cardinality of the table.

公开(公告)号：CA2317081C

公开(公告)日：2004-06-01

申请号：CA2317081

申请日：2000-08-28

Applicant: IBM CANADA

Inventor： HAAS PETER JAY ,  LOHMAN GUY MARING ,  RJAIBI WALID

IPC: G06F17/30

Abstract: Column cardinality for a table in a partitioned relational database is estimated. In each node in the database containing partitioned portions of the table a hashing function is carried out to generate a hash data set for each node, according to one of sample counting, linear counting or logarithmic counting techniques. The hash data sets are merged in a coordinator node to generate a combined hash data set. The combined hash data set is counted and correcte d according to the appropriate sample counting, linear counting or logarithmic counting techniques. The resulting count is adjusted for hash collisions and the resultant value is used to estimate the column cardinality of the table.