一种广域网中的分布式互斥算法

An Algorithm for Distributed Mutual Exclusion in WAN

广域网可简单地看作由多个局域网通过远程通信线路互连组成。为了适应广域网环境的特点，文章提出了一种两层结构的分布式互斥算法模型，把广域网系统组织成由局部进程组成的局部网络和由每个局部网络中的协调进程组成的全局环两层。为了互斥地访问共享资源，局部进程必须首先获得局部令牌，然后再向本地协调进程申请全局令牌，只有获得了局部和全局令牌的局部进程才能进入临界区。还讨论了对该算法可能的扩展。

WAN can be simply taken as a system comprised of some interconnected LANs, which contact with each other through remote communication. In order to adapt the communication environment in WAN, an algorithm for distributed mutual exclusion is presented in this paper. This algorithm organizes WAN into a logic two-level model, one level of which is local networks comprising of local processes, the other level of which is a global ring made up of every coordinator of every local network. According to this algorithm, in order to access a shared resource exclusively. a local process must get the local token at first, and then requests global token to local coordinator. Only a local process that get both the local and global token can enter critical section. Possible extensions to the algorithm are also discussed in this paper.