Data Management in P2P Systems.
Carlos Eduardo Santos Pires, Bernadette Farias Lóscio, Ana Carolina Salgado · Brazilian Symposium on Databases · 2006
Data Grids are an innovative sweet spot taking advantage of existing computer science concepts in file systems, database systems and grid computing. Data Grid Management Systems (DGMS), through multiple abstraction mechanisms provide a logical namespace (or logical view) that hides the complexity of distributed data and heterogeneous resources. The logical namespace provided by DGMS is independent of physical organization of data or storage resources. DGMS middleware are used to implement data grids, digital libraries, persistent archives, and dataflow systems to manage unstructured data. Managing unstructured data that is distributed in multiple countries using traditional file systems is not feasible. DGMS seems to satisfy in this technical requirement felt by global companies and large scientific projects. Currently several petabytes of unstructured data in the form of files are managed around the world using DGMS technologies. A DGMS provides a schema to organize the data based on storage attributes, domain-specific metadata, user access permissions and so on. Based on the current usage patterns in production scenarios, several open research problems can be observed. Event-driven data management based on ECA rules is one such promising research area for automation of DGMS.