A caching model for efficient distributed network and systems management

This paper is motivated by the increasing need for scalable, distributed management architectures for integrated network, system and application management. Such integration and extension of management functionality impose heavy performance requirements, in direct conflict with the increased volume of management traffic produced. We present an attempt to minimize this traffic and overall response time, by applying the concept of “quasi-copies” (initially introduced in information retrieval technology) in the management information collection process. The proposed scheme improves overall performance of distributed management architectures by taking advantage of various consistency requirements, inherent in distributed management applications. Components are organized in a distributed hierarchy of “management information” clients (managers) and servers (agents) or dual role entities. In an attempt to identify the expected performance gains, we present a simple queuing model that provides analytical results on expected improvement of response time and reduction of management traffic