Article Preview
TopIntroduction
Cloud computing creates new business models and value chains, changing the way computing, storage and networking resources are purchased and consumed (Buyya, Yeo, Venugopal, Broberg, & Brandic, 2009). In the cloud ecosystem, more people and organizations benefit from the ubiquitous computing access. The cloud aims to deliver infrastructure (IaaS), platform (PaaS), software (SaaS), data (DaaS), network (NaaS) and everything related to computing as a service under the term Everything as a Service (XaaS) (Banerjee et al., 2011). The services are consumed either independently or combining two or more services. Cloud services are offered in different standards, management interfaces, pricing schema, quality of service (QoS) and resource availability options. In this diverse market, search and selection of the best set of suitable services to match the requirements of a consumer becomes challenging.
The cloud marketplace addresses current cloud market hurdles, involving complex information processing, and limitation in flexibility of searching and selecting the required cloud solutions from multiple vendors. The marketplace allows multiple cloud service providers (CSPs) to offer resources to consumers in one place, where consumers can select a wide range of service from diverse providers (infrastructure providers, platform providers, application providers, service resellers, intermediators, service aggregators and service enablers). The marketplace enables consistency and standardization of services with unified billing and one-stop cloud solutions for consumers (Menychtas et al., 2012).
Selection of appropriate cloud services based on consumers' requirement is an important aspect of the marketplace in order to allow the elasticity that is required at business level. Cloud marketplaces should be capable of supporting service compositions, spanning the cloud stack layers, and provide mechanisms and tools to the developers for reusing existing services. This requires automated selection processes not only for services but also for pricing models and service level agreements (SLAs) through which these cloud services are provisioned. Thus, the service selection becomes a very complicated process, especially when cloud marketplaces are capable of offering any kind of cloud services (XaaS), each of which can be merchandised through different pricing schemas and SLAs (Menychtas, Gatzioura, & Varvarigou, 2011). Selection of appropriate software packages not only improves the efficiency of work, but also can increases the related service revenues (Jadhav, & Sonar, 2009).
The Cloud Services Measurement Initiative Consortium (CSMIC) (CSMIC, 2014) developed Service Measurement Index (SMI) with critical characteristics, associated attributes and measures that provide a standardized method for measuring and comparing current non-cloud based services with cloud based services or cloud services available from multiple providers. The SMI is designed to measure critical business and technical requirements of any type of cloud services. In this paper, a framework (CloudSelect) has been proposed to compare and select the best suitable cloud services in the cloud marketplace based on identified functional and non-functional attributes of SMI. The CloudSelect allows consumers to compare different cloud services based on their needs, prorates and with other criteria.
The process of evaluation and selection of the cloud services involve simultaneous consideration of multiple criteria to rank the available alternatives and select the best one. It requires functional and non-functional requirements of the consumer and is a Multi criteria decision making (MCDM) problem, which refers to making preference decisions over the available alternatives that are characterized by multiple, usually conflicting criteria (Tzeng, & Huang, 2011). The goal of the MCDM is: (i) to help decision makers to choose the best alternative, (ii) to outrank the alternatives that seem to be good among the set of available alternatives, and (iii) to rank the alternatives in decreasing order of their performance (Mansooreh, & Pet-Edwards, 1997).