The SRB is a client-server middleware that uses datasets to build a logical name space for referencing distributed data.
The SRB provides a way to organize information stored on multiple heterogeneous systems into logical entities for ease of use.
The SRB provides access to data stored on archival resources, file systems and databases.
The SRB also provides capabilities to store replicas of data, controlling and auditing access to documents and collections, which is also known by logical-to-physical mapping hence Metadata Cataloging
The SRB also provides a facility for co-locating data together using containers with more flexibility in accessing and updating files and stores user-defined metadata at different levels for better search capabilities.
What is SRB?
SRB specific capabilities
Support for the execution of remote operations.
Access controls and audit trails to control and track data usage
When a system failure occurs that would bring down an MCAT fragment, but shouldn’t affect the entire SRB, as the other sections will isolate the temporarily inaccessible server.
Federation permits MetaData to reside closer to the resource
To reduce network latency, an MCAT server is hosted closer to the resource; in turn the SRB performance improves on a network span basis.
Federation enables the Virtual Organizations notion and makes it easier to merge newbies to the network, not to forget the ability of transferring and replicating documents across VOs
Cons & Pros
Cons
That paper can be considered as a fine support for understanding the Data Grids Architecture.
Experiencing such enormous projects describes the powerful and huge capabilities of an SRB.
