ABSTRACT -- At the High Performance Computing Laboratory (HPCL), cluster computing (also known as Beowulf cluster) systems are used to reduce the workload of the existing supercomputers. The scope of the cluster system in this paper is to build the system which can efficiently handle parallel and sequential jobs. Furthermore, this system is designed to house the large disk capacity ( bytes or one Terabyte) by using the software RAID (Redundancy Array of Inexpensive Disk). Using RAID, we can improve the performance of the read and write I/O bottlenecks. Here, we present both architectural aspect and software that make up this cluster computing system. The I/O performance issue will also be discussed.

KEYWORDS -- Cluster Computing, RAID, Storage Device, Server, Parallel Computation