Fuzzy Cloud Service Selection Framework Masoumeh Tajvidi Department of Computer Engineering, Shahid Beheshti University,Tehran, Iran m.tajvidi@webmail.sbu.ac.ir Rajiv Ranjan CSIRO Computational Informatics, Canberra, Australia rajiv.ranjan@csiro.au Joanna Kolodziej Institute of Computer Science, Cracow University of Technology, Cracow, Poland jokolodziej@pk.edu.pl Lizhe Wang Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing , China Wanglz@ radi.ac.cn Abstract— Cloud computing is the latest computing paradigm that delivers hardware and software resources as virtualization enabled services. Recently, cloud service selection has emerged as an important research problem due to large number of cloud providers and their diverse service configurations, multiple selection criteria, and customer’s fuzzy perception of Quality of Service (QoS). In this paper, we propose a novel fuzzy logic framework for cloud service selection based on the individual QoS criteria of customers. In this model, the necessary service configuration and run-time QoS data is collected from reliable sources such as monitoring services, customers' feedbacks, and certified cloud providers’ information. The obtained results from conducted case study shows validity and applicability of the proposed framework. Keywords — Fuzzy Logic; Cloud Service Selection; Cloud Provider; Multi-Criteria Decision Process; Monitoring tools; User feedback. I. INTRODUCTION Cloud computing paradigm [1], [2] provides users a wide range of services based on a flexible pay-as-you-go pricing model. The flexibility of dynamically acquiring (based on application workload demands) cloud services have encouraged many enterprises to migrate their IT applications to cloud-based virtualized services (e.g. CPU, storage, network, databases, application server, and web server). Fuelled by such demands leading IT vendors including Google, Microsoft, and Amazon has started offering variety of cloud services. Naturally, it is challenging [3] for customers to select the right mix of cloud services that meet their QoS criteria at different stages of application lifecycle orchestration (e.g. selection, deployment, and run-time management). The first step in migrating application to the cloud is to select the best mix of cloud services that can deliver the QoS agreed as part of Service Level Agreement (SLA). However, selecting QoS optimized cloud services is not a trivial task due the heterogeneities of service configuration naming [4] and diverse QoS features. For instance, a low-end CPU service of Microsoft Azure is 30% more expensive than the comparable Amazon EC2 CPU resource, but it can process application workload twice as quickly. One of the available approaches for choosing cloud services is to manually read the configuration information from provider’s websites. However, such an approach does not support streamline and easy-to-understand comparison of cloud service configurations, specially based on mix of often conflicting QoS criteria. Further the information published by providers is not 100% reliable, as the providers may exaggerate the claims about their services’ capabilities in order to overthrow other competitors [5],[6]. Moreover, it is extremely daunting task for customers to directly map their application QoS needs (e.g. maximize web server throughput) to the service configuration (e.g. available RAM on a CPU resource or available processing power of CPU resource) published by providers. The importance of service selection issue returns to the fact that migrating from one cloud provider to another is not only costly but also a risky process. If the selection process is not optimally executed it could lead to vendor lock-in issues such as dependency of application stack on particular virtualization format (e.g., Hyper-V, Xen, KVM, etc.) or programming platforms (Amazon SimpleDB, Microsoft SQL Server, etc.). Even though the cloud federation has been coined to solve some of the issues mentioned above, and the technical issues involved with establishing such a federation are numerous. Hence, cloud federation is not likely to become reality in near future. It is clear that cloud service selection should be carefully undertaken while considering customer’s QoS needs and available services’ configuration. In order to assist customers with the selection process and enable them to select the most appropriate (as per QoS needs) service, a reliable cloud service selection framework is required. To this end, we propose a novel fuzzy logic framework that undertakes service selection based on the individual QoS criteria of customers. Although many research proposed models and frameworks for ranking cloud services exist, to the best of our knowledge, most of the existing approaches neglect configuration and QoS information capture and validation phase. So the existing approaches are biased with uncertain information of could service providers, consumer’s vague conception of the requirement, and also unrealistic measurement of QoS which depends on real time measurement and past QoS history of the services [4]. The main contributions of this framework can be described as follows: • Ability to validate the captured configuration service information via a third party validator • Ability to obtain realistic run-time measurement of QoS attributes from monitoring tools • Ability to include customers feedback information (about past performance of services) in the selection process • Ability to simplify the customer’s fuzzy perception of QoS by modeling the problem as a fuzzy multi criterion decision-making process. The proposed approach can 2014 IEEE 3rd International Conference on Cloud Networking (CloudNet) 978-1-4799-2730-2/14/$31.00 ©2014 IEEE 443