Spatio-temporal data services in a shared-nothing environment
Marios Hadjieleftheriou, Vassil Kriakov, Yangui Tao, George Kollios, Athanasios Delis, Vassilis J. Tsotras · 2004
Recently, there has been a proliferation of applications creating spatio-temporal data that has to be processed, stored and queried efficiently. Existing applications may produce Terabytes of raw data per day that routinely necessitate the execution of millions of update operations, so as to keep the underlying database up-to-date. Consequently, there is a need for spatio-temporal data management systems that are able to support such update intensive operations. Moreover, these systems should offer users the capability to examine past and present versions of the data in an on-line fashion. In this respect, we propose a system that exploits the inherent parallelism of a shared-nothing computing environment for storing and indexing the spatio-temporal data. Our infrastructure consists of a cluster of workstations (COW) connected via a Gigabit/sec network, with servers whose operation is based on a distributed multi-version indexing scheme. We describe our proposed system architecture, data organization, as well as pertinent algorithms. We discuss various optimizations whose objective is to ensure robustness and scalability under highly dynamic situations manifested by excessive query loads and high update rates. Finally, initial experimental results using a system prototype and simulated environments support the effectiveness of our approach.