Sampling for database systems

摘要: A database server supports weighted and unweighted sampling of records or tuples in accordance with desired semantics such as replacement (WR), without (WoR), independent coin flips (CF) semantics, for example. The may perform sequentially not only to sample non-materialized records, those produced a stream by pipeline query tree example, but also whether materialized not, single pass. over join two relations requiring the computation full materialization both and/or indexes on attribute values relations.