Collaborative Load-Balancing in Storage Networks Using Agent Negotiation

The rapid advances in the computer industry allow building larger systems that require mass storage volumes. As storage space increases, its management becomes increasingly difficult. In contemporary enterprise storage systems, performance has developed into a major bottleneck, which negatively affects the overall IT performance. Centralized solutions are often infeasible and thus a distributed solution should be sought. Our novel approach involves incorporating intelligent agents to the storage system, allowing the utilization of a distributed negotiation scheme between agents that act on behalf of the clients who require storage and on behalf of the storage servers within the system. Using a simulation environment which models real settings of a large storage network, we demonstrate the benefits of applying our distributed solution in the storage management domain in terms of client satisfaction, servers' revenue, and overall system performance. Our results show a significant improvement in storage performance when this solution is implemented.