分布式超级节点选举算法

摘要

基于超级节点的分布式系统中,若超级节点失效或临时离开,希望系统能够自组织地选举出能力最强的节点作为新的超级节点.提出分布式超级节点选举算法,通过洪泛过程构造底层的生成树,叶子节点沿此树进行消息的传递,消息中包含着关于节点和边的信息,根节点根据这些信息构造最小生成树.根节点选出能力最强的节点作为超级节点,并沿着最小生成树广播选举结果.对算法性能从通信复杂度和时间复杂度两方面进行了分析和比较.%In super-node-based distributed systems,if super-node failures or temporarily leaves,it is hoped that the most capable node could be elected as new super-node in self-organization way. This paper presents distributed super-node election algorithm. It firstly constructs bottom spanning tree through process of flooding,and then leaf nodes pass messages along this tree.Information about nodes and edges is contained in the messages, which are used by root node to construct minimum spanning tree(MST).At last, root node selects the most capable node as new super-node and broadcasts election results along MST. The algorithm performance is analyzed and compared from aspects of communication complexity and time complexity.