Browse Theses

The type of the workload on a database management system (DBMS) is a key consideration in tuning its performance. Allocations for resources such as main memory can be very different depending on whether the workload type is Online Transaction Processing (OLTP) or Decision Support System (DSS). A DBMS also typically experiences changes in the type of workload it handles during its normal processing cycle. Database administrators must, therefore, recognize the significant shifts of workload type that demand reconfiguring the system in order to maintain acceptable levels of performance. We envision autonomous, self-tuning DBMSs that have the capability to manage their own performance by automatically recognizing the workload type and predicting its change over time. In this thesis, we make two main contributions to the development of autonomic DBMSs. The first contribution is a methodology for automatically identifying a DBMS workload as either OLTP or DSS by building various classification models. We demonstrate the methodology with both industry standard workloads and with real workloads of global financial firms. The second contribution is a prediction architecture to forecast when the type of a workload may change. The DBMS can therefore proactively adjust its parameters, without incurring the overhead associated with the constant monitoring. We present experiments to show that the performance of the DBMS using our prediction mode outperforms other possible operation modes. They also show that the prediction architecture can adapt to changes in the workload pattern. The architecture does not demand human intervention and is potentially a generic solution for other similar prediction problems.