Fault-tolerant resource placement in balanced hypercubes

Abstract

Balanced hypercube is a variant of the standard hypercube structure for multicomputers, with desirable properties of strong connectivity, regularity, and symmetry. This structure is a special type of load balancing graph designed to tolerate processor failure. In a balanced hypercube, each processor has a backup (matching) processor that shares the same set of neighboring nodes. Therefore, tasks that run on a faulty processor can be reactivated in the backup processor to provide efficient system reconfiguration. In this paper, we study the resource placement problem in balanced hypercubes. Resources can be hardware devices (such as I/O processors, disks, etc.) or software modules (such as data files, library routines, etc.). The proposed placement algorithm guarantees that each node without resource copy connects two (matching) nodes with resource copy. Therefore, one of the resource copies at each matching pair can be used as a backup to tolerate possible processor failures. Moreover, the proposed algorithm is perfect in that each node without resource copy connects two and only two nodes with resource copy and neighbors of each node with resource copy are nodes without resource copy.