AQuES: an agent-based query evaluation system

Abstract

Optimization for distributed and parallel database systems is an overly complex problem. This complexity arises due to the various kinds of parallelism and the coexistence of heterogeneous hardware modules. The query optimizers for those systems must exploit parallel algorithms for algebra operators, independent data parallelism and resource scheduling. In addition, unexpected deviations of the workload at runtime enforce new, dynamic execution strategies. This paper reports ongoing work in developing an agent-based query optimization and execution system to cope with this problem. It is based on the principles of distributed, symmetric processes which work in a cooperative, yet independent manner. This approach enables a dynamical adaptation to changes in the execution environment with local and simple actions. We present the architecture of the components and outline the principle of distributed query execution within this software system.