Efficient management of massive data sets is a key aspect in typical grid and e-science applications. To this end, the benefits of employing database technologies in such applications has been identified since the early days of grid computing, which aims at enabling coordinated resource sharing, knowledge generation and problem solving in dynamic, multi-institutional virtual organizations through a distributed, scalable, adaptive and autonomous infrastructure. Nowadays, databases are playing an increasingly important role, both when the data is (semi-) structured and when it is stored in flat files. This survey paper discusses in detail existing database-oriented grid middleware. It focuses on several complementary aspects, such as dynamicity, autonomy, resilience to failures, and performance, and presents the characteristics, capabilities and limitations of existing solutions.
展开▼
