Adaptive Data Management in Evolving Heterogeneous Hardware/Software Systems-II
Our aim is to develop new processing concepts for exploiting the special characteristics of hardware accelerators in heterogeneous system architectures for classical and non-classical database systems. On the system management level, we want to research alternative query modeling concepts and mapping approaches that are better suited to capture the extended feature sets of heterogeneous hardware/software systems. On the hardware level, we will work on how processing engines for non-classical database systems can benefit from heterogeneous hardware and in which way processing engines mapped across device boundaries may provide benefits for query optimization. Our working hypothesis is that standard query mapping approaches with their consideration of queries on the level of individual operators is not sufficient to explore the extended processing features of heterogeneous system architectures. In the same way, implementing a complete operator on an individual device does not seem to be optimal to exploit heterogeneous systems. We base these claims on our results from the first project phase where we developed the ADAMANT architecture allowing a plug & play integration of heterogeneous hardware accelerators. We will extend ADAMANT by the proposed processing approaches in the second project phase and focus on how to utilize the extended feature sets of heterogeneous systems rather than how to set such systems up.