Small Materialized Aggregates (SMAs for short) are considered a highly flexible and versatile alternative for materialized data cubes. The basic idea is to compute many aggregate values for small to medium-sized buckets of tuples. These aggregates are then used to speed up query processing. We present the general idea and present an application of SMAs to the TPC-D benchmark. We show that exploiting SMAs for TPC-D Query 1 results in a speed up of two orders of magnitude. Then, we investigate the problem of query processing in the presence of SMAs. Last, we briefly discuss some further tuning possibilities for SMAs.
展开▼
