[CALCITE-344] Lattice data structure

• Type: New Feature
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.9.1-incubating
• Assignee: Julian Hyde
• Reporter: Julian Hyde
• Created: 13/Jul/14 22:11
• Updated: 18/Oct/14 22:36
• Resolved: 29/Aug/14 18:11

Description

Add a "lattice" data structure, to organize, gather statistics on, and recommend materialized queries.

DDL. Here is a possible SQL DDL syntax:

CREATE LATTICE SalesStar AS
SELECT *
FROM SalesFact AS s
JOIN TimeDim AS t USING (timeId)
JOIN CustomerDim AS c USING (customerId)

Structure. A materialized query would optionally belong to a lattice.

Constraints. Creating a lattice implies that the first table is the fact table of a star schema, and all joins are many-to-one. That is, foreign keys, primary keys and NOT NULL constraints are implied.

Statistics. When queries arrive that could use a lattice, a counter is incremented.

Recommendation. An agent could recommend materialized queries to create based on static analysis of the star schema (e.g. table and column cardinalities), and perhaps statistics of past use.

View matching. The optimizer uses a lattice to identify materialized queries that could satisfy a query. (Without a lattice, the such space would be much larger, because the optimizer would have to consider many join permutations.)

Issue Links

• relates to: CALCITE-402 Lattice should create materializations on demand