Research Initiation: Using Multicast Communication for Resource Finding in Distributed Systems

本研究解决了分布式系统中资源查找的一般问题。前提是，在需要快速更新位置信息或系统包含大量节点的系统中，无法在不造成重大性能损失的情况下使用该问题的现有解决方案。研究了利用组播通信降低资源查找成本的方法。在广播网络中，要探索的算法将利用在许多网络接口中为组播提供的硬件支持。在点对点网络中，任务是形式化搜索算法的概念，并研究当资源位置的概率分布被定义时，如何找到最小化平均带宽或时间的搜索算法。研究了在点对点网络中实现组播通信的问题。许多用于查找远程对象的算法将在云分布式操作系统中实现。

This research address the general problem of resource finding in distributed systems. The premise is that existing solutions to this problem cannot be used, without a substantial performance penalty, in systems where the location information needs to be updated rapidly or when the system contains a large number of nodes. The use of multicast communication to reduce the cost of resource finding is investigated. In broadcast networks, the algorithms to be explored will make use of the hardware support provided for multicast in many network interfaces. In point-to-point networks, the task is to formalize the notion of a search algorithm and to study the problem of finding search algorithms that minimize the average bandwidth or time when a probability distribution is defined for the location of a resource. The problem of implementing multicast communication in point-to-point networks is studied. Many of the algorithms for finding remote objects will be implemented in the Clouds distributed operating system.