On Construction of a Large Computing Farm Using Multiple Linux PC Clusters

Abstract

In addition to the traditional massively parallel computers, distributed workstation clusters now play an important role in scientific computing, due to the advent of commodity high performance processors, low-latency/high-bandwidth networks and powerful development tools. In a cluster environment, we add more and more workstations for reaching more computing performance. Unfortunately, in general case, we don’t have enough space for setting up many PCs to form a large-scale PC cluster, and more questions we are concerned about, for example, network topology, electric power, and management question etc. Thus, we can use several places for setting up a number of machines on each place, and enabling networking for transparent computing on each cluster node for enable an enormous node cluster. In this paper, we use some routing policy and a kernel module named Netfilter to enable three or more clusters to be connected as a big one. In our experimental environment we use three different quantities of nodes in each PC cluster connected via a Fast-Ethernet network environment. As experiment, we performed matrix multiplication for performance analysis.

This work is supported by National Center for High-Performance Computing (NCHC), Taiwan under Grant No. NCHC-KING_010200.