PPT Slide
Horizontal partitioning approach
- N separate physical tables with:
- Object id column
- Columns for every subtable column (inherited or not)
So what did we do for UDB V5.2...?
- Vertical partitioning approach rejected quickly
- Too many joins to materialize subtable rows
- Multi-column constraints and indices problematic
- Horizontal partitioning approach rejected eventually
- Uniqueness issue for user-generated object ids
- Query complexity for multi-hierarchy join queries
- Ex: select p.name, q.name from Person p, Person q where ...
- Implementation table approach taken for V5.2
- Appeared to give us the most "free" functionality
- Adopted despite row size (null columns) downside
Implementing Table Hierarchies (cont.)