File-based replica management

Erasure Codes for Increasing the Availability of Grid Data Storage

“…Clearly the justification to choose our approach to store a file is dependent on the expected time to store, and the frequency of file access. If a very fast and frequent access to files are required a replication infrastructure [10] can be used in addition to our application.…”

“…To minimize data access latencies automatic replication and replica selection frameworks have been developed. By taking into account the networking monitoring data during automatic replica selection the access time in wide area transfers can be significantly reduced [10]. Other work on Grid data management has studied the coupling between data access and the scheduling of jobs.…”

An implementation of parallel file distribution in an agent hierarchy

“…For high file availability, Coda facilitates disconnected operations in a file system by allowing mobile clients to hoard files, emulate servers, and reintegrate files into their original servers [25]. For lower file-access latency, the Globus replica-management architecture has implemented an LDAP-based replica catalog API that allows users to select a given replica [26]. The EU DataGrid project has even automated this replica selection that locates the best replica with regard to network and storage access latencies [27].…”

RRS: Replica Registration Service for Data Grids

“…In the European Data Grid Project a similar replica management framework was implemented [7]. An integrated approach for data and meta-data management is provided in the Storage Resource Broker (SRB) [1].…”