A new heartbeat mechanism for large-scale cluster

Abstract

Distributed managed clusters have appeared in recent years, and computing intensive scientific problems request large-scale clusters. However, many of the traditional heartbeat mechanisms do not fit large-scale distributed managed clusters. In this paper, we propose a switch-based heartbeat mechanism named heartbeat ring, which adapts to large-scale distributed managed clusters. Heartbeat ring mechanism has the prominent advantages in simplicity, scalability and adaptability, and so on. Finally, based on a prototype implemented on Linux platform, experiment evaluation is presented. © Springer-Verlag Berlin Heidelberg 2006.