The BigDawg monitoring framework

BigDAWG is a polystore database system designed to work with heterogenous data that may be stored in disparate database and storage engines. A central component of the BigDAWG polystore system is the ability to submit queries that may be executed in different data engines. This paper presents a monitoring framework for the BigDawg federated database system which maintains performance information on benchmark queries. As environmental conditions change, the monitoring framework updates existing performance information to match current conditions. Using this information, the monitoring system can determine the optimal query execution plan for similar incoming queries. We also describe a series of test queries that were run to assess whether the system correctly determines the optimal plans for such queries.