QUERY OPTIMIZATION OVER DISTRIBUTED HETEROGENEOUS EXECUTION ENGINES

摘要

A system includes reception of a database query, determination of a first logical query execution plan to execute the database query over a plurality of heterogeneous execution engines, selection of a first logical operator of a first operation level of the first logical query execution plan, identification of a first one or more physical operators corresponding to the first logical operator and an output format of each of the first one or more physical operators, each of the first one or more physical operators provided by a respective one of the plurality of heterogeneous distributed execution engines, selection of a second logical operator of a second operation level of the first logical query execution plan, the second logical operator to receive output from the first logical operator, identification of a second one or more physical operators corresponding to the second logical operator, each of the second one or more physical operators provided by a respective one of the plurality of heterogeneous execution engines, determination of a first resource usage estimate for each of the first one or more physical operators, determination of one or more second resource usage estimates for each of the second one or more physical operators, based on the respective output format of each of the first one or more physical operators, and determination of one of the first one or more physical operators and one of the second one or more physical operators based on the determined first resource usage estimates and the determined second resource usage estimates.