Efficient Resource Allocation and Migration ...

Jour of Adv Research in Dynamical & Control Systems, Vol. 9, No. 6, 2017

Efficient Resource Allocation and Migration Management Using Distributed Computing J. Surya Kumar, Research Scholar, Dept of Computer Science, Vels University, Chennai. E-mail:[email protected] K. Sharmila, Assistant Professor, Dept of Computer Science, Vels University, Chennai. E-mail:[email protected] Dr.G. Suseendran, Assistant Professor, Department of Information Technology, Vels University, Chennai. E-mail:[email protected]

Abstract--- Cloud computing is a very large server in which different services and data are stored and can access by all. The software and data that we want to access doesn’t exist on our computer instead it’s on the server. Similarly the cloud computing acts as a larger server. The service is divided into three categories. They are SAAS, PAAS, and IAAS. It provides Information Technology as a service over the Internet with delivery on demand and payment based on usage. Previously improper use of Virtual Machines leads to the imbalance load distribution and an increase in the operation cost. The problem identified for the study is to minimize the amount of extra resources reserved on each PM during server consolidation and to reduce the live migrations of VMs for a better performance of cloud computing. Now it is proposed to use each Virtual Machine as a two-state Markov chain to capture burstiness; then it is designed as a resource reservation strategy for each physical machine based on stationary distribution of a mark chain. This migration of Virtual Machine provides a method to distribute physical resource more reasonably without stopping service, so that less energy is consumed and operation cost is reduced. In this research paper how Hot machines and Warm machines are used to assign automatic migration and the implementation is discussed. Keywords--- Cloud Computing, Virtual Machine, Hot Machine, Warm Machine.

I.

Introduction

CLOUD computing has been gaining more importance in the past few years and it is changing the way of access and retrieve information. The recent emergence of virtual desktop has further elevated the importance of cloud computing. Cloud is a very large server in which different services and data are stored and any one can access all those for any work. Normally, the software or data, that any one wants to access for the work, does not exist in the computer itself instead they get from the server. This concept of using services from server, not stored on the system, is called Cloud Computing which is available in three kinds of services like Infrastructure as a service (IAAS), Platform as a service (PAAS) and Software as a service(SAAS). IAAS as a service provided by providers like AWS, supplies a virtual service instance and storage, as well as application program interface (API) that user send workload to virtual machines. The user allocates to the storage capacity and can use start, stop access and configure the VM storage as desired. In PAAS model, provider offers tools on their infrastructure and the user accepts these tools over the internet using API as web portals or gateway software. SAAS is a distribution model that delivers software applications over the internet. As a crucial technique in modern computing clouds, virtualization enables physical machine (PM) to host many performance-isolated virtual machines (VMs). It greatly helps a computing cloud where VMs running various applications are aggregated together to improve resource utilization. In the earlier works the cost of energy consumption like power supply and cooling occupied a significant fraction of the total operating costs in a cloud. Therefore, making optimal utilization of underlying resources to reduce the energy consumption is becoming an important issue. To cut the energy consumption in clouds, server consolidation is very essential and it is possible by which tightly packed VMs reduce the number of running PMs; however, VMs’ performance may be seriously affected if VMs are not appropriately placed, especially in a highly consolidated cloud. The variability and burstiness of VM workload widely exists in modern computing clouds.

II.

Proposed Work

The proposed system architecture consist of two main sub-systems. Local resizing and live migration are the two pervasively-used methods. Local resizing adjusts VM configuration. On the other hand, live migration moves some VM(s) to a relatively idle PM, when local resizing is not able to allocate enough resources. However, in a highly consolidated computing cloud where resource contention is generally prominent among VMs, live migration may

ISSN 1943-023X

169


cause significant service downtime; furthermore, it also incurs noticeable CPU usage on the host PM, which probably degrades the co-located VMs’ performance. Thus, it is of great importance to reduce the energy consumption and operating cost, and to enhance the efficiency of data centers by proper use of VMs. At present, three primary measures can be employed to improve the efficiency of VMs. They are (1) Improving the framework and distribution by enhancing the security and computing efficiency of VMs by improving either the framework or by optimal distribution of VMs. (2) Scheduling tasks in a Cloud as tasks and distribute them in a way similar to tasks scheduling in traditional distributed computing to reduce energy consumption. (3) Migrating VMs dynamically: Dynamic migration of VMs provides a method to distribute physical resource more reasonably without stopping the service. Therefore, in the present study it is proposed to reserve some extra resources on each PM to accommodate bursty workload. It also investigates the problem of minimizing the amount of extra resources reserved on each PM during server consolidation. Thus the problem can be formulated for optimization leading to minimize the amount of resource reserved on each PM, tasks scheduling in distributed computing to reduce energy consumption and to reduce number of live migrations of VMs by which overall performance of a computing cloud could be improved.

III.

Methodology

In the proposed work, each VM as a two-state Markov chain to capture burstiness, design a resource reservation strategy for each physical machine based on stationary distribution of a markov chain. This migration of VM provides a method to distribute physical resource more reasonably without stopping service, so that less energy is consumed and operation cost is reduced. The modification process is the implementation process in the present study. In this process two types of systems are deployed; Hot Machine to handle the current job and Warm Machine kept in idle state until job is assigned. Three Virtual servers are deployed for every machine. First Job is assigned to the first Virtual machine and in the same way jobs assigned to other VMs in the Hot machine. Now jobs are assigned to the Warm machines once all the VMs of Hot category have occupied with the jobs. Automatic migration of job is implemented, so as to transfer the load to the Hot VM from Warm VM once it has completed the job. Cache mechanism is also implemented in the present study. The modification process is having some advantages like avoiding congestion, less power consumption, decreased waiting time, and high data transmission rate, reliable and avoid replicate request.

There are many modules to the proposed work. They are discussed below User Registration In this module, a User application is created by which the User is allowed to access the data from the Server of the Cloud Service Provider. Here, first the User wants to create an account and then the user is allowed to access the Network. Once the User created an account, login the account and request the Job from the Cloud Service Provider. Based on the User’s request, the Cloud Service Provider will process the User requested Job and respond to them. All the User details will be stored in the Database of the Cloud Service Provider. In this Project, the user designs the User Interface Frame to communicate with the Cloud Server through Network Coding using the programming

ISSN 1943-023X

170


Languages like Java/ .Net. By sending the request to Cloud Server Provider, the User can access the requested data if they authenticated by the Cloud Service Provider. Cloud Server Deployments Cloud Service Provider contains the large amount of data in their Data Storage. The Cloud Service provider will maintain all the User information to authenticate the User when login into their account. The User information will be stored in the Database of the Cloud Service Provider. Also the Cloud Server will redirect the User requested job to the Resource Assigning Module to process the User requested Job. The Request of all the Users will be processed by the Resource Assigning Module. To communicate with the Client with the other modules of the Cloud Network, the Cloud Server will establish connection between them. For this Purpose we are going to create a User Interface Frame. Also the Cloud Service Provider will send the User Job request to the Resource Assign Module in Fist in First out (FIFO) manner. Intermediate Server Deployments By implementing Intermediate Server the Job Processing Scheme will effectively process the User Requested Job and efficiently maintains the Resources of the Cloud Server, so that we can save the Energy of the Resources when they are not processing the Job. Green Computing Deployments Green computing is the term used to denote efficient use of resources in computing. It is also known as Green IT . In this Module, the User requested Job is processed. The User requested Job will redirect to the RAM of the Cloud Server. The RAM contains three Types of the Physical Servers namely the HOT Server, the WARM Server and the COLD Server. These Physical Servers will contain ‘n’ number of virtual Servers to process the User requested Job so that the Job can be efficiently processed. Migration of Virtual Server In this module the migration server is created. The main use of migration is to migrate the job from one virtual server to another server so that the energy can be reduced and work load of the server is balanced. By using the Migration we can shift the process from one VM to anther VM without loss of data. Cache Server Implementation Creating a Cache Memory in the User requested job will store for a period of time. If another User requests the same Job to the Server of the Cloud Service Provider (CSP), the Server will check in the Cache Memory first so that we can reduce the Job Processing Time. Then the Server will provide the Data to the User immediately. If the request Data is not in the Cache Memory, then the Server will process the User requested Job by transferring it to the RAM.

Architecture

IV. S.NO 1 2 3 4 5 6 7 8 9

Algorithm Used in Previous Work AUTHOUR Moreno Marzolla, Ozlap Babaoglu, Fabio Panzieri. Norman Bobroff ,Andrzej Kochut, Kirk Beaty Srikanth Kandula , Sudipta Sengupta, Albert Greenberg, Praveen patel, Ronnie chaiken Ningfang Mi , Giuliaro castale, Ludmila cherkasova Meng Wang, Xiaoqiao meng, Li zhang Dixie, Ning Ding ,y.charlie hu, Ramana Kompella Sheng Zhang, Zhuzhong Qian, Jie Wu, sanglu Lu Fahimeh Farahankian, Paris Liljeberg, Tapio, Juha Gasto keller, Hanan Lutfiyya

ISSN 1943-023X

ALGORITHM Active and Passive threads MFR algorithm TASK sheduling algorithm MAP_Sample algorithm Group packing algorithm TIVC Allocation algorithm I-DSEN Allocation algorithm VM Consolidation MS-ES Relocation policy algorithm

171


V.

Results and Discussions

In this study each Virtual Machine as a two-state Markov chain to capture burstiness, was designed then a resource reservation strategy for each physical machine based on stationary distribution of a markov chain. This migration of VM provides a method to distribute physical resource more reasonably without stopping service, so that energy is consumed and operation cost is reduced. In this implementation process two types of deploy systems are used namely Hot Machines which can handle the current job and Warm Machines which are kept idle state until job is assigned. Thus for every machine three Virtual servers are deployed. Firstly Job is assigned to the Hot machine of first Virtual machine and same way following jobs are assigned to other Virtual Machine. Now jobs are assigned to the Warm machines once all the VMs of Hot category have occupied with the jobs. Automatic migration of job is implemented, so as to transfer the load to the Hot VM from Warm VM once it has completed the job. Here four physical machine are used with each has three virtual machine. Once the login process is completed it requests the source files and opens the allocated physical system(jobs). After the allocation has completed the ninth job the tenth job is migrated toVM1 and then to VM2, a process called migration . Jobs are migrated but also the ranges of bytes values are sequentially over, so the PM4 allocation is completed. It just transfers PM1 in VM1 and maximized the results.

ISSN 1943-023X

172


The source file in the available format is present in different kinds of available resources. The user can retrieve some required file format which can be moved to the selecting file format. Once it is completed, the downloaded event is migrated in the files list and it can be seen visually in the output screen. The feasibility of the project is analyzed in this phase and business proposal is put forth with a very general plan for the project and some cost estimates. During system analysis the feasibility study of the proposed system is to be carried out. This is to ensure that the proposed system is not a burden to the company. For feasibility analysis, some understanding of the major requirements for the system is essential. Three key considerations involved in the feasibility analysis are • • •

Economical Feasibility Technical Feasibility Operational Feasibility

Economical Feasibility This study is carried out to check the economic impact that the system will have on the organization. The amount of fund that the company can pour into the research and development of the system is limited. The expenditures must be justified. Thus the developed system as well within the budget and this was achieved because most of the technologies used are freely available. Only the customized products had to be purchased. Technical Feasibility This study is carried out to check the technical feasibility, that is, the technical requirements of the system. Any system developed must not have a high demand on the available technical resources. This will lead to high demands on the available technical resources. This will lead to high demands being placed on the client. The developed system must have a modest requirement, as only minimal or null changes are required for implementing this system. Operational Feasibility The aspect of study is to check the level of acceptance of the system by the user. This includes the process of training the user to use the system efficiently. The user must not feel threatened by the system, instead must accept it as a necessity. The level of acceptance by the users solely depends on the methods that are employed to educate the user about the system and to make him familiar with it. His level of confidence must be raised so that he is also able to make some constructive criticism, which is welcomed, as he is the final user of the system.

VI.

Conclusion

In a highly consolidated computing cloud, the VM performance is prone to degradation without an appropriate VM placement strategy, if various and distinct burstiness exists. To alleviate this problem more PMs are activated, leading to more energy consumption. To balance the performance and energy consumption with respect to bursty workload, the present study proposed to reserve a certain amount of resources on each PM that forms a queuing system to accommodate burstiness. To quantify the amount of reserved resources is not a trivial problem. In this study, a burstiness-aware server consolidation algorithm based on the two-state Markov chain has been proposed. We studied the probabilistic performance constraint and showed that the proposed algorithm is able to guarantee this performance constraint. Ultimately the required green computing job is achieved.

VII.

Future Enhancement

Possible future research topics for the dynamic migration problem in the cloud computing platform, which includes, the generation, learning and mutation method of Bucket Code, can be further improved to reduce its asymmetry property. One example is calculating proper probability for generating the first part of the Bucket Code, proposing a better way for codes to learn from each other instead of just learn from the best one.

References [1] [2]

Armbrust, M., Fox, A., Griffith, R., Joseph, A.D., Katz, R., Konwinski, A., Lee, G., Patterson, D., Rabkin, A., Stoica, I. and Zaharia, M. A view of cloud computing. Commun. ACM 53 (4) (2010) 50–58. Oh, M.H., Kim, S.W., Kim, D.W. and Kim, S.W. Method and architecture for virtual desktop service. U.S. Patent 20 130 007737, 2013.

ISSN 1943-023X

173


[3] [4] [5] [6] [7] [8]

Marzolla, M., Babaoglu, O. and Panzieri, F. Server consolidation in clouds through gossiping. Proc. IEEE Int. Symp. World Wireless, Mobile Multimedia Netw., 1–6. Vogels, W. Beyond server consolidation. ACM Queue 6 (1) (2008) 20–26. Bobroff, N., Kochut, A. and Beaty, K. Dynamic placement of virtual machines for managing SLA violations. Proc. IFIP/IEEE Int. Symp. Integr. Netw. Manag., 2007, 119–128. Kandula, S., Sengupta, S., Greenberg, A., Patel, P. and Chaiken, R. The nature of data center traffic: Measurements & analysis. Proc. ACM 9th ACM SIGCOMM Conf. Internet Meas. Conf., 2009, 202–208. Mi, N., Casale, G., Cherkasova, L. and Smirni, E. Injecting realistic burstiness to a traditional client-server benchmark. Proc. IEEE 6th Int. Conf. Auton. Comput., 2009, 149–158. Xie, D., Ding, N., Hu, Y.C. and Kompella, R. The only constant is change: Incorporating time-varying network reservations in data centers. Proc. ACM SIGCOMM, 2012, 199–210.

ISSN 1943-023X

174