Infrastructure Utilization in Cloud Computing to Increase Provider Revenue

...

performance reduction. This SLA gives good suggestion about increasing revenue but they are rejecting the proposal which workload is higher then their capacity and they are also rejecting some running SLAs after finding some profitable SLA. These steps will create doubtful situation in business [7].

Linlin Wu, Saurabh Kumar Garg and Rajkumar Byya (2011) worked on SLA and algorithm to maximize the revenue of SaaS provider. The main objective of this work is to maximize the profit for provider by minimizing the costs of virtual machines. By providing individual VM to every users then no QoS degradation will occur so no penalties and more number of customers. SaaS Company increases their profit by providing new VM to new request and upgraded VM to upgraded requests. The main focus of this work is on the dynamic changing requests of customers. Software provider tries to increase their profit and to ensure QoS to expand their business. Software is delivered in standard, professional and enterprise. Costumer’s accounts are created in group, team and department. Contract is signed between Software provider and customer. If any one violate then defaulter pay his penalties. This work is good step to increase the revenue of Software provider but the main focus of this research work is only SaaS. There is no alternative way if customers request load is higher than their resources capacity. So they will reject such like proposal which have very bed effects on business[8].

Hong Xu and Bachun Li (2011) worked on revenue maximization model to increase the revenue of the Cloud computing provider. For better utilization of computing resources and to increase the revenue of provider now a days dynamic pricing mechanism is used for dynamic customer’s requests. Amazon EC2 is also offering dynamic pricing sine 2009. Dynamic pricing mechanism in IaaS causing too many problems.They formulated a program which deal with such like problem and handle infinite horizon cases[9].

Qi Zang, ErenGijrses ,Raouf Boutaba and Jin Xiao (2012) developed a framework to maximize the revenue of cloud provider. They are using market analyzer which analyzes the market incoming request briefly. Secondly they are using capacity planner which prepare the machines and resources capacity according to the reports of market analyzer. This model is using both price mechanism, dynamic and static price mechanism. Different algorithms are used to predict the situation to use suitable price mechanism technique which mechanism would be suitable for certain situations. [10].

3. Proposed Model

Three types of parties are participating in this proposed model. One is IaaS Provider party which is providing his physical resources. Second is consumer party which hires the resources from IaaS provider party. Third is external cloud service. IaaS hire some resources from third party to minimize SLA violation and to give reliable services to the consumers.

www.theinternationaljournal.org > RJSITM: Volume: 04, Number: 03, January-2015 Page 40

Figure 4: Proposed IaaS Provider Model

Two types of SLAs are established in this model. One SLA is between IaaS provider and consumers and second SLA is between IaaS and external Cloud Services. The End users can dynamically change their request and there is no need to upgrade account. They will be charged according to their usage of resources. IaaS will give resources according to their dynamic request. IaaS does not hire complete resources from external cloud service but it pay to external Cloud services according to the usage of service. External cloud provider charged the IaaS provider with CostTotal=Cost Perunit * t, so we will try to use external resources only in emergency to minimize the external payments.

Algorithm 1: Pseudo Code

1. Comp_resources

2. Hired_resources

3. Sort(Sort VMs according to their available resources)

4. Customer_request(request_resources)

5. If(Deploy the VM to the Job using first fit policy){

6. If(VM has enough resources){

7. Put job on that VM

8. }

9. }

10. If(Utilization of VM increase from its available resources){

11. Shift the task to furtherVM having enough resources

12. }

13. If(request_resources increases from Comp_resouces){

14. Shift the portion of this job to external cloud services

15. }

16. If(Comp_resources is enough to run portion of the job running on hired resourcs){

17. Copy it back to Comp_resources

18. }

In algorithm 1 Comp_Resource shows the IaaS provider own physical infrastructure. Hired_Resources show those resources which are hired from any external Cloud provider. The working of this model is described in steps.

1. All resources (VMs) available with the IaaS provider are sorted according to their capacity.

2. Proposed IaaS model receive request from customers. Request is checked with SLA.

3. All VMs are already sorted. VM is allotted to the job with first fit mechanism. If the VM have enough space then job is put on it.

4. If the utilization of the current VM is increasing from its capacity then job is shifted to any other machine using algorithm steps from 3 to 9.

www.theinternationaljournal.org > RJSITM: Volume: 04, Number: 03, January-2015 Page 41

5. If the requested job workload is higher than the company own resources then hired resources are utilized for that job to give the customer reliable service. External resources are only used in emergency. Job is shifted back to company own resources when they become free to minimize the external payments and to minimize the company own resources.

6. If any job is running on external hired cloud and there is enough capacity to run that job on company resources then job is shifted back to company own resources.

4. Experimental Design

We used CloudSim to simulate proposed IaaS model. In simulation experiment we created n VM as company resources, n VM as hired resources

...