Cloud Service Matchmaking using Constraint Programming

TL;DR

This paper presents a constraint programming-based approach for cloud service matchmaking that effectively handles various QoS property types and user preferences, improving upon existing discrete-value methods.

Contribution

It introduces a new constraint model addressing list-typed QoS properties and explicit preference handling, advancing service matching techniques with constraint solvers.

Findings

Constraint solvers effectively handle soft constraints in service matching.

The approach covers all analyzed QoS property types.

Prototype demonstrates feasibility and identifies implementation challenges.

Abstract

Service requesters with limited technical knowledge should be able to compare services based on their quality of service (QoS) requirements in cloud service marketplaces. Existing service matching approaches focus on QoS requirements as discrete numeric values and intervals. The analysis of existing research on non-functional properties reveals two improvement opportunities: list-typed QoS properties as well as explicit handling of preferences for lower or higher property values. We develop a concept and constraint models for a service matcher which contributes to existing approaches by addressing these issues using constraint solvers. The prototype uses an API at the standardisation stage and discovers implementation challenges. This paper concludes that constraint solvers provide a valuable tool to solve the service matching problem with soft constraints and are capable of covering all QoS property types in our analysis. Our approach is to be further investigated in the application context of cloud federations.