Architecture and Implementation of Distributed Data Storage Using Web Services, CORBA and PVM

摘要

We propose an architecture and its implementation called PVMWebCluster I/O targeted for distributed data storage and retrieval. Data is submitted via Web services from geographically distant clients or via CORBA from within clusters which offers better performance. The system consists of many, possibly geographically distributed clusters which run DAMPVM/PVM. Each cluster has a manager waiting for CORBA read/write calls which are passed to particular nodes in the cluster via PVM messages. Each cluster has a corresponding Web service that passes read/write calls to the CORBA manager. Web services form a top layer of PVMWebCluster I/O and call each other to obtain the best cluster/node to store a particular chunk of data. This results in a very flexible architecture which integrates distributed clusters for data storage of small and large files. The architecture proposes caching at both Web service and cluster layers. We have tested an initial implementation by submission of files of varying sizes in four configurations: via the Web service and CORBA layers on a local machine, via Web services in a LAN and via Web services through the Internet.