HPC: A Hierarchical Parameter Configuration Approach for Failure Detection Scheme in Application Layer Multicast
Jijun Cao, Zhigang Sun, Jinshu Su · 2009
Failure detection is the basis of recovery process of partitioned application layer multicast (ALM) tree caused by node failures. However, the parameter configuration issues, which directly affect the performance of failure detection scheme, have seldom been concerned. By analyzing the disadvantage of traditional basic parameter configuration (BPC) approach for probe-based failure detection scheme in ALM, we first define a new metric, i.e. Detection Loss, to measure the performance of various parameter configuration approaches in this paper. And then we propose a hierarchical parameter configuration (HPC) approach, in which the configurable parameters of failure detecting processes in each monitor node are configured mainly according to the level in which its target node is located in ALM tree. The theoretical analysis and numerical results show that the HPC approach can not only balance the detection loss of various nodes in ALM tree but also reduce the detection loss of ALM tree without any increased overall control overhead as compared to the BPC approach. Finally, some possible directions for the future work are discussed.