Service Level Agreement Complexity: Processing Concerns for Standalone and Aggregate SLAs

TL;DR

This paper analyzes the complexity of managing multiple service level agreements in cloud systems, proposing a model for machine-readable SLAs and examining the computational challenges involved.

Contribution

It introduces a simple model for machine-readable SLAs, discusses their construction for cloud infrastructure, and characterizes the computational complexity of SLA policy injection.

Findings

Injecting SLA policies into cloud infrastructure is NP-Complete in the general case.

Constraining SLA representations can make the problem more tractable.

Approximation techniques can be used to address computational complexity.

Abstract

In this paper, we examine the problem of a single provider offering multiple types of service level agreements, and the implications thereof. In doing so, we propose a simple model for machine-readable service level agreements (SLAs) and outline specifically how these machine-readable SLAs can be constructed and injected into cloud infrastructures - important for next-generation cloud systems as well as customers. We then computationally characterize the problem, establishing the importance of both verification and solution, showing that in the general case injecting policies into cloud infrastructure is NP-Complete, though the problem can be made more tractable by further constraining SLA representations and using approximation techniques.