Late Materialization
A query optimization strategy in which only lightweight metadata is read during the filtering and join stages, and full column data is fetched only after the candidate set has been reduced, thereby lowering I/O and memory bandwidth pressure.
Timeline (last 90 days)
Late materialization is a query optimization strategy: during the filtering and join stages, only lightweight metadata (such as bounding boxes and row numbers) is read, and once the candidate set is narrowed down, the full column data is read back from the table.
The three technologies of vectorized execution, late materialization and spatially aware data layout are often used in combination, achieving significant synergistic acceleration effects on lakehouse columnar storage.
All Facts (2)
Late materialization is a query optimization strategy: during the filtering and join stages, only lightweight metadata (such as bounding boxes and row numbers) is read, and once the candidate set is narrowed down, the full column data is read back from the table.
50%UnverifiedThe three technologies of vectorized execution, late materialization and spatially aware data layout are often used in combination, achieving significant synergistic acceleration effects on lakehouse columnar storage.
50%