Infrastructure Utilization in Cloud Computing to Increase Provider Revenue

www.theinternationaljournal.org > RJSITM: Volume: 04, Number: 03, January-2015 Page 37

SLA based Infrastructure resources allocation in Cloud computing to increase IaaS provider revenue

Ateeqa Jalal

NCBA Chenab College Mianwali, Pakistan

&

Afzal Badshah Khattak

Education Department, Punjab, Pakistan

&

Tauseef-U-Rehman

Faculty of Technology, Preston University Islamabad-Pakistan

Abstract

Cloud Computing is an emerging technology in which desktop services are delivered online on network.Consumers pay for the service how much they use such as water, electricity telephone or gas.Service consumer does not need to invest capital amount on infrastructure and IT experts.

Revenue is the main focus of any business. In Cloud Computing revenue depends on proper VMs utilization and SLA.Cloud infrastructure provider have their own resources or they hire some resources to meet the customers requirements. Hiring other infrastructure, cost too much and performance depends on other hired systems which lead towards SLA violation. SLA violation in Cloud business minimizes the revenue of the Cloud provider.

This paper discusses the IaaS model with SLA to maximize the Infrastructure provider revenue. Two types of SLAs are implemented in this model. One SLA is between service consumer and provider and second SLA is between hired external resources and IaaS provider. We propose such polices and algorithm which increase the company own infrastructure utilization, minimize the response time and SLA violation to enrich the revenue. This algorithm and policies will reduce the infrastructure cost and minimize the SLA violation. It will handle the dynamic changes in the customer request and will allot resources with dynamic pricing according to their dynamic requests.

Keywords: Cloud Computing, Infrastructure as a Service, Service Level Agreement,

Infrastructure Resources Allocation, Revenue Maximization

1. Introduction

Cloud Computing is becoming very important for business. Instead of buying infrastructure, operator, license and software, customers simply hire cloud services on cheaper rates[1]. Cloud Computing is not a new technology but it is the combination of old technologies. Cloud Computing is self-descriptive and self-corrective computer services. Cloud Computing use virtualization approach to provide different services. It is extremelysuccessful paradigm for utility computing [2][3].

Figure 1: Cloud Computing

www.theinternationaljournal.org > RJSITM: Volume: 04, Number: 03, January-2015 Page 38

Cloud Computing provides three types of services. Software as a Service, provide online application such as application, monitoring finance and communication etc, Platform as a Service, provide online development tools such as testing, analysis and deployment service. Infrastructure as a Service provides online physical resources such as computing, storage and networking etc[3].

Cloud Computing services are provided in three model. Private Cloud resources are visible only for one organization users. Public Cloud computing resources can be hired by any person after getting registration. Hybrid cloud computing is the combination of private and public cloud computing [4].

Infrastructure as a Service provides physical resources of computer online on network such as computing, storage and bandwidth etc. First of all Infrastructure as a Service cost too much secondly it have too many problems such as configuration ,management and maintenance etc [1].

Figure 2: Infrastructure as a Service

Therefore small and medium level organization can’t invest too much capital investment on IaaS at initial stages of business. They simply hire IaaS services to start their business [2].

Service Level Agreement is contract negotiated between service provider and service consumer. The need of SLA is to increase trust between both parties.SLA provides benefits such as it enhances customer satisfaction in term of Quality of Service and improvesrelationship between two parties. Penalties are imposed on failure party [4].

Figure 3: Service Level Agreement

In cloud computing three types of pricing mechanisms are used. Tiered pricing in which Cloud computing services are divided in different tires.each tire have different prices. Amazon use tiered pricing scheme. Unit pricing in which customers have been charged on unit of space or bandwidth

www.theinternationaljournal.org > RJSITM: Volume: 04, Number: 03, January-2015 Page 39

used.This pricing mechanism is more flexible than tired pricing mechanism. Subscription base pricing in which customers are charged according to their subscription [4].

Cloud Computing is very emerging and hot topic in industry and academic. 1950 damp terminal were connected with main frame computer, so resources sharing idea was exist that time too. In 1960 Mr. John Mclarthy gives an idea that computing resources can be delivered at home as other utility resources [5]. In 1970 the concept of virtualization has come. VM machines share one physical resource for too many users.Virtualization is the core of Cloud Computing [6]. In 1990 telecommunication companies started virtual private network, before that dedicated lines were used for each customer. In1999 salesforce.com was the first Cloud services provider. Now days Amazon is the leading Cloud Computingservices provider. Google also has stared his Cloud business in 2009 [5][6].

2. Related Work

Mario Macfas, J.OriolFito and JordiGuitart (2003) worked on revenue maximization in cloud computing using EERM. This used bidirectional data between market and resources to increase revenue. They used dynamic pricing mechanisms. Price will be higher if the workload is higher. When provider is unable to fulfill customers request, SLA violation list is created then those SLA are violated which cause lower lose.If new customer is coming with new SLA which profit is more than one other SLA penalties then the older SLA is cancelled. There are different QoS SLAs. Combination of these SLAs are adopted and those are violated or cancelled which cause lower lose. If one machine is overloaded then some more VMs are taken to reduce the