Group communication as an infrastructure for distributed system management

Abstract

In the past, system management tools for computer systems were oriented towards managing a single computer with, possibly, many users. When the networked system concept became widespread, centralized solutions such as the Network Information Service (NIS) were developed to help the system manager to control a network of workstations. Today, when many sites contain hundreds of workstations, these solutions may no longer be adequate. This paper proposes the usage of techniques, developed for group communication and database replication, for distributed cluster management. We show how group communication can be exploited to provide three classes of frequently needed operations: simultaneous execution of the same operation in a group of workstations; software installation in multiple workstations; and consistent network table management (improving the consistency of NIS).