Dynamic service configurations for SLA negotiation

Abstract

Utility Computing based infrastructures such as Cloud Computing promise on-demand packaging of resources similar to metered public utilities, i.e. electricity, water, gas and telephone. However, computing resources, which are traded as services are very different from the usual commodities due to their dynamically changing behavior and (re)configurable properties. Service Level Agreements(SLA) ensure the necessary guarantees to the highly dependent service consumers. There may be several rounds of negotiation before a formal SLA is established. During automated negotiation sessions, the service provider needs to understand consumer requests and is required to offer the closest possible service configuration fulfilling these requirements, keeping in view the preferences of the consumer on one hand and the business rules and configuration constraints of the service provider on the other. Service providers need to be prepared to expect demands for all possible permutation of service attributes. This requires a mechanism to map given expected values onto a discrete or continuous set of possibilities and then their refinement through multiple negotiation rounds. In this paper, we present a formal approach to compute feasible configurations of services, which fulfill the consumer preferences as well as the service provider's constraints and then introduce an SLA negotiation process based on this formal model. © 2011 Springer-Verlag Berlin Heidelberg.