A Robust and Efficient MCDM-Based Framework for Cloud Service Selection Using Modified TOPSIS

Introduction

Cloud computing plays a vital role for start-ups as well as small enterprises to start their business with low capital investment. It is a new form of utility computing that delivers computing resources, storage, software, platform, etc. on-demand to the customers over the internet on pay as per usage like other utilities (e.g. gas, Internet data, electricity, etc.). It also provides scalable, ubiquitous, reliable and dynamic services to customers to fulfill their requirements. Its services are broadly categorized into three major areas- Software as a Service (SaaS), Platform as a Service (PaaS) and Infrastructure as a Service (IaaS). SaaS provides ready-made applications for use like customer relationship management (CRM) (Cusumano, 2010), Gmail, Saleforce.com, etc. PaaS provides an environment or platform like Google App Engine (Ciurana, 2009), Windows Azure Cloud Services to develop applications on top of it. IaaS provides compute, network, and storage services for hosting the applications. The major IaaS service providers are Amazon Web Services (AWS), IBM Bluemix, Rackspace, Google Compute Engine, etc.

Cloud computing is boon for medium and small enterprises as they do not have to invest a large amount of capital on their infrastructure setup and maintenance. They have to only focus on their innovation and business policy as computing resources, storage, network, platforms, etc. are available at a competitive cost on a subscription basis and resources are maintained by the cloud service providers (CSPs). It also provides the freedom to the business community from scalability and elasticity (Shojafar, 2016) perspective as services are scalable within a minute based on demand. Due to the above benefits, many companies like Coursera, Siemens, Netflix, etc. (AWS Customer Success, 2018) are managing and developing their services on the cloud. It is also motivating large enterprises like Google, IBM, Microsoft, etc. to move their business to cloud due to its accessibility at anytime and anyplace. Its usage is growing at a very fast rate as its users are from the individual level to small enterprises to big organizations. But there are a lot of challenges to move applications to cloud as each application has specific requirements like response time, security, availability, etc. So, before moving the applications to the cloud, there is a need to select the best CSP that could fulfill the specific requirement of the application.

As there is a scope of a lot of profit in cloud computing in upcoming future due to its growing use, many big organizations like Microsoft, Google, IBM, Amazon, etc. have invested huge capital and started providing cloud services. Therefore, for a cloud user, it has become a difficult task to choose an appropriate CSP available in the market like AWS, IBM Bluemix, Google Cloud, Salesforce, Microsoft Azure, etc. according to their requirements as they are providing similar QoS at different cost and performance. There are many CSPs that provide one service like computing at lower cost and storage at a higher cost than the others. Some CSPs also charge a different cost for a service based on regions like AWS EC2 (Varia, 2011) has different pricing for EC2 instances based on the region where it is located. All the CSPs are competing based on QoS provided by them.

So, the selection of right CSP as per cloud user requirement is a challenging task as there are many CSPs in the market and its selection depends on many functional and non-functional QoS requirements. Therefore, there is a need for a framework that can assist the cloud user in efficient and accurate way to select the right CSP and rank them. In this context, there is a need for identification of standard QoS metrics for cloud services and a ranking algorithm to rank CSPs based on QoS. To identify the standard QoS metrics, the leading global organizations of the world formed a consortium named Cloud Services Measurements Initiative Consortium (CSMIC) (SMI framework 2019) to identify the QoS metrics for services provided by CSPs. CSMIC proposed a framework called Service Measurement Index (SMI), which has seven major QoS metrics like Accountability, Financial, Usability, Performance, Agility, Assurance, and Security & Privacy. Further, they divided each metric into various sub-metrics to describe each one in detail. These metrics are used by cloud users to compare various CSPs.