IUP Publications Online
Home About IUP Magazines Journals Books Archives
     
Recommend    |    Subscriber Services    |    Feedback    |     Subscribe Online
 
The IUP Journal of Information Technology
An Improved Throttled Virtual Machine Load Balancer for Cloud
:
:
:
:
:
:
:
:
:
 
 
 
 
 
 

With the growing popularity of cloud computing today, its model will one day serve as the fifth utility service outside electricity, telephone, water and gas. Visualizing how this utility model will revolutionize the way people use computing resources requires that we take the big challenge of performance unpredictability associated with load imbalance, computing resource distribution inefficiency and minimum resource consumption into consideration. There is a need for developing a better model that will not only reduce cost but also make enterprise as per userís satisfaction. Improving resource utility and performance of distributed system in a way that will yield better response time, processing time and efficient virtual machine monitoring is therefore, the concern of this research work. Throttled load balancing algorithm was analyzed and its deficiencies serve as a basis for improvement in the proposed system. The proposed system rearranges Virtual Machines (VMs) according to their threshold value and 80% threshold value for each machine serving as the maximum utilization range for cloudlets allocation. The proposed system then spreads load across all VMs until each machine attains 80% of its threshold value. If this level is attained and there are still cloudlets at the global queue, the 20% un-utilized threshold value can then be used. The system also monitors VM efficiency and stops allocation to any VM that does not perform to its optimum level. An extensive simulation was carried out to evaluate the proposed system using Cloud Analyst simulator in order to compare the existing system and the proposed system. The results show that the proposed system yields a better response time and lower turnaround time and provides efficient VM monitoring than the existing throttled load balancer.

 
 

With the advent of the era of infrastructure-less computing, user community has started exploring options to move from traditional infrastructure investment to outsourcing infrastructure deployment based on the utility model (Ansuyia and Deepak, 2013). With this development, the research community has realized the need to move from the much-hyped utility computing models to a more realistic cloud. This improvement has resulted in the philosophies of infrastructure-less and utility computing, which has evolved a new computing paradigm called cloud computing (Ansuyia and Deepak, 2013). Cloud computing can be compared with early proliferation of electricity, where homes, businesses and towns find it difficult to produce or rely on their source of power and rather they wish to connect to a greater power grid usually supported by utilities (Eva, 2010). Using this utility connection help saves time and cost, provides greater access to and more reliable availability of power. The term cloud computing has been defined by many researchers. Foster et al. (2008) described cloud computing as a delivery of computing as a service rather than product with shared resources, software and information provided to computers and other devices as a utility (like electricity grid) over a network (Foster et al., 2008). Buyya et al. (2008) defined cloud computing as a type of parallel and distributed system with collection of interconnected and virtualized computers, which are provisioned dynamically and presented as one or more unified computing resources based on Service Level Agreements (SLA) established through negotiation between service providers and cloud users (Buyya et al., 2008). Cloud computing has been accepted and adopted by many industries, but so many issues about this new utility model has not been completely addressed. Some of these issues include load balancing, virtual machine migration to prevent against failure, server consolidation, energy management to prevent carbon emission which is dangerous to health, and reduction in cost (Deepak, 2014).

Most important concern in all these issues is load balancing in cloud, which is a mechanism to distribute workload evenly in cloud to achieve a better user satisfaction and resource utilization ratio (Deepak, 2014). With the bottleneck of load imbalance, computing resource distribution inefficiency and minimum resource consumption, there is a need for developing a better model that will not only help in reducing costs but also make enterprises as per user satisfaction. Improving resource utility and performance of distributed system in such a way that will reduce response and processing time with better virtual machine efficiency policy is, therefore, the major concern of this research study. The aim of this paper is to develop an improved throttled load balancer that will improve response and turnaround time for all usersí requests and results of the enhanced system compared with the existing throttled load balancing algorithm in cloud analyst simulator.

The rest of this work is organized as follows. Section 2 gives a description of cloud computing, load balancing and review existing load balancing algorithms in cloud. Section 3 gives a detailed description of the proposed algorithm. Section 4 describes the experimental setup and the simulation parameter configuration of the improved system, comparative analysis of the proposed system and the existing system, followed by a discussion in Section 5. Finally, the paper ends with conclusion.

 
 

Information Technology Journal,Virtual machine, Datacenter, Virtualization, Threshold value, Cloudlet, Userbase, Cloud analyst.