Replicated Data Management in the Grid: The Re:GRIDiT Approach

Grid environments more and more target novel domains suchas eScience, eHealth or digital libraries that feature a varietyof data-intensive applications. Consequently, issues relatedto data management in Grids are becoming increasingly important.In terms of data management, the Grid allowskeeping a large number of replicas of data objects, possiblywith different versions or levels of freshness, to allow for ahigh degree of availability, reliability and performance so asto best meet the needs of users and applications. At thesame time, the seamless integration of replication managementinto the Grid while taking into account its special characteristics,needs to be done without any central componentfor managing data or metadata. In this paper, we report onthe ongoing Re:GRIDiT project which aims at addressingall the above requirements. Re:GRIDiT distinguishes betweenpotentially many updateable and read-only replicaswhich can be distributed across a Grid environment. First,Re:GRIDiT provides new protocols for the correct synchronizationof concurrent updates to different updateable replicasand their subsequent propagation in a completely distributedway. Second, Re:GRIDiT takes into account thesemantics of the data which is managed in the Grid: mutabledata can be subject to updates; immutable data, in turn,cannot be changed once created, but may be subject to versioncontrol. Third, Re:GRIDiT will be dynamic in a waythat according to the current load, new replicas (updateableor read-only) can be created or removed on demand. Fourth, Re:GRIDiT will provide read-only transactions the full flexibilityto specify the freshness (for mutable data) or versionnumber (for immutable data) – which is particularly usefulin order to trade accuracy for performance in the access todata in the Grid.