System and method for incrementally maintaining non-distributive aggregate functions in a relational database

A computer system for incrementally maintaining non-distributive aggregate functions in materialized views of relational databases, comprising:

• means for selectively altering a predetermined propagate phase data structure to yield a new propagate phase data structure by
• (a) building an AST query graph;
• (b) building a join predicate between a predetermined propagate phase data structure and said AST query graph;
• (c) pushing said join predicate down to each leaf operation of said AST query graph;
• (d) adding said join predicate to each leaf operation where said join predicate is to the top of said a predetermined propagate phase data structure;
• and
• means for selectively applying changes to at least one materialized view according to said new propagate phase data structure.

A computer program product comprising a machine-readable medium having machine-executable instructions thereon including code for incrementally maintaining non-distributive aggregate functions in materialized views of relational databases, said code comprising:

• a first code for selectively altering a predetermined propagate phase data structure to yield a new propagate phase data structure;
• and
• a second code for selectively applying changes to at least one materialized view according to said new propagate phase data structure, wherein said altering occurs only if all of the following conditions are false;
• (a) all functions of a query are distributive;
• (b) deltas in said query are insert-only, and said functions are either distributive or MAX functions or MIN functions;
• and
• (c) only deletes need be applied to said materialized views.

A method for incrementally maintaining non-distributive aggregate functions in materialized views of relational databases, comprising:

• selectively altering a predetermined propagate phase data structure to yield a new propagate phase data structure;
• selectively applying changes to at least one materialized view according to said new propagate phase data structure, wherein said applying includes:
• if any inserts to said materialized view are required, then building a left outer join between said new propagate phase data structure and said materialized view, else building an inner join between said materialized view and said new propagate phase data structure;
• if any updates to said materialized view are required, then adding an update operation on said materialized view at the top of said new propagate phase data structure with the corresponding predicate and iteratively building set clauses for all aggregate functions;
• if any deletes to said materialized view are required, then adding a delete operation on said materialized view at the top of said new propagate phase data structure with the corresponding predicate;
• if any inserts to said materialized view are required, then adding an insert operation on said materialized view at the top of said new propagate phase data structure with the corresponding predicate.