Scheduling and controlling semantics for distributed resource based computing engines

Abstract

With the advent of autonomic and cloud computing, computation engines are getting redefined as dynamic configurations of heterogeneous, distributed resources. In this paper, we describe the operational semantics of scheduling and controlling of computation engines configured from component resources subject to dependency and capacity constraints and in accordance with policies and objectives such as priorities and load balancing. The operational semantics provides a novel formal model in denotational style, for establishing properties like computability and dependability in the presence of faults and reported and unreported events. It supports dynamic features such as resource up and down events, synchronized startup, synchronized shutdown, and resource groups/virtual servers. An efficient, interpreter-based implementation using the specified semantics is suggested. © 2009 IEEE.