A Survey Paper on Enhancing Infrastructure-as-a-Service to the Needs of Performance in VM using Cloud

Download Full-Text PDF Cite this Publication

Text Only Version

A Survey Paper on Enhancing Infrastructure-as-a-Service to the Needs of Performance in VM using Cloud

Kavyashree B L Prof.Nandhish A C

M.tech 2nd sem , CSE Dept.of.CSE

City Engineering College City Engineering College Bangalore. Bangalore.

Abstract —Cloud environment aims to produce the level of configuration and deployment requirements. Cloud also involves the requisition interaction between the user and service provider. Cloud addresses the use of virtual machines as a single set of physical resources. The performance in virtual machines is a long term traces from Amazon web services ,Google app engine. This investigation of performance using cloud environment exhibits yearly and former pattern levels of applications. The impact of performance is assessed through simulation using some applications such as execution of job, trading of goods in virtual networks. Many cloud service provider are underlying to provide large number of services to the clients is a major problem faced. To overcome this problem, this paper has been presented to enhance the IaaS to the needs of performance utilization.

Keywords – Resource allocation, cloud computing, performance.

  1. INTRODUCTION

    Cloud computing is an emerging technology that provides efficient use of resources for different services which are cost effective[2]. An infrastructure is provided for computational environment in the form of virtual machines undertaken by cloud services. The performance is stable relating to the scalability of cloud services regarding to the conceptual tasks that have been carried out by virtual machines. The infrastructure service is based on the performance variability assessed by some of the cloud service providers such as Amazon and Google[3].

    Cloud aims to provide the deployed applications on demand of cost effectiveness based on the quality of service that the user provides. In order to improve resource utilization a single set of physical virtual machine must be replaced by several sets of virtual machines which is stable in nature. According to many researchers the cloud services such as Infrastructure-as-a-service(IaaS),Platform-as-a- service(PaaS),Software-as-a-service(SaaS) provides a better performance variability in virtual machines. Some developers bring out new ideas in implementing services using software and hardware infrastructures[1]. The stability of performance in many real time operating systems is critical. In order to provide a better performance

    this paper is presented to analyze the efficiency utilization in virtual machines using cloud services.

    AWS

    AWS

    The challenge of quantifying the performance analysis and better understanding and evaluation of some of the sheduling policies is based on aspects like system size, requirements of Qos based on cloud.

    5

    4.5

    4

    3.5

    3

    VM

    5

    4.5

    4

    3.5

    3

    VM

    2.5

    2.5

    1

    0.5

    0

    1

    0.5

    0

    Stability Reliability scale

    Stability Reliability scale

    2

    1.5

    2

    1.5

    Google

    Google

    Fig.1.Chart representing performance metrics

    Virtual machine is compared with the cloud service providers such as AWS and Google for the enhancement of performance metrics given in Fig.1.

  2. RELATED WORKS

    1. Deployment Service

      An infrastructure service to be deployed , job should be set- up. Some of the minor issues occur such as VM sprawl , Memory shortages , Storage Shortages , and performance bottlenecks. Utilization and responsiveness are defined in regarding to the set of performance. Some of the model are scalable in representing resources in both physical and virtual manner in order to exploit this infrastructure-as-a- Service. In order to set-up a VM in an infrastructure an agreement(SLA) should be signed up between the clients and the service provider. The VM's are not affected by signing to the SLA indeed they provide a better

      performance metrics in order to carry out the major tasks in a least cost and with the better time consumption.

    2. Elasticity Computing

      In order to provide a better elasticity, Amazon Web Services provides a platform to deliver the services to the requested clients by the IaaS delivery model. AWS involves the actions carried out by Elastic compute cloud to provide web hosting environments. Some of the environments like simple queue service(SQS), simple DB, Elastic Block Storage(EBS), Simple Storage Service(S3), Virtual Private Cloud(VPC) all undergo in the AWS management console.

      Elastic Compute Cloud is a web service involved in launching instances with a simple interface which consists of some applications that run under several Operating Systems such as Linux Distributions, Microsoft Windows Server 2003 and 2008, OpenSolaris, Free BSD and NetBSD.

      • Maximum capacity carried by VM over a single pool of Physical machines.

    The correlation between the service provider and the clients is identified by the resource utilization and performance characteristics. It is the fundamental technology that powers cloud computing.

  3. SYSTEM MODELS

    A. Provisioning of Virtual and Physical machines

    Virtual machine instances are created using pre-built image by the request which is processed by IaaS. Some of the request specific instances are CPU, RAM, ROM and disk capacity. The deployment of Virtual machines are based on the number of physical machines that are acting on one another. Each of the physical machines are shared by the multiple number of virtual machines[4]. Physical machines must be running in order to reduce the time delays and accuracy in VM provisioning. These must be turned-on but should not be ready for further communication between the virtual machine and the

    AWS

    EC2

    EC2

    Free BSD Open

    service provider.

    10

    8

    6

    4 VM

    2 PM

    EBS

    solaris 0

    Fig 2. Representation of AWS and EC2

    1. Prediction of Virtual Machine

      The resource utilization improvement from an infrastructure service provider dealing with the performance cost is variably undertaken by the Virtual Machine.To meet a performance utilization carried out by virtual machine there must be a strong determination of physical node to get the appropriate resources. Prediction of performance in virtual machine is done by resource manager to co-ordinate the activities of the applications that are running in physical nodes.

      The probability of job rejection reduces by increase in the capacity of the pool of physical machine. The infrastructure service in virtual machine to provide better performance analysis must include some of the characteristics, they are:

      • Count of Physical Machine.

      • Large number of virtual machine corresponding with equal number of physical machine.

      • Minimum service time that is carried out by a job recommendation.

      • Average time to calculate the delay in VM's.

    Fig 3. Graph representing efficiency of VM

    B. Administering of VM infrastructure

    The efficiency and consistency should be in large number so that the host profiles allow administrators to set-up the host systems. The templates remain constant without changing in current environment.VM should be administered, rebooted and repaired in order to provide a better efficiency. The elasticity is proved by releasing and provisioning of the capabilities perhaps the demand can be accessed in both inward and outward manner.The cloud infrastructure was a asset in earlier days, but it is a service in the current era. Cloud infrastructure is not a business but it provides a social service to the requested clients in order to get the efficient way of utilization of the host systems. Therefore this paper provides the ability to enhance the cloud infrastructure being used in Virtual machine. Essentially, virtualization differs from cloud computing because virtualization is software that

    manipulates hardware, while cloud computing refers to a service that results from that manipulation.

    This paper summarizes valuable information and practical steps for IT managers who want to plan and implement private cloud infrastructure as a service (IaaS) as the first step toward cloud services delivery, including:

    • How building a cloud service delivery model will help your organization take full advantage of the agility and efficiency benefits of cloud computing

    • The key technologies and capabilities that you need to move from an IT virtualization practice to a private cloud computing practice

    • A framework for approaching your private cloud project that lays the groundwork for moving to a hybrid model when you are ready

    • A quick review of the five leading cloud management platforms (CMPs): Apache, CloudStack, Eucalyptus, Cloud platform, Microsoft, cloud software, OpenStack, cloud software, and VMware, vCloud Director.

  4. PERFORMANCE ANALYTICS

    1. Load balancing

      This effect of load balancing is very important in the performance of cloud infrastructure. In order to provide too many servers and virtual machine in an organization it is better to provide a single server within a cloud infrastructure. Many clients access the requested services from the server through the services offered by the infrastructure. In order to increase the efficiency and reliability two or more virtual machines can be replaced by a single set of virtual machine within a cloud. Therefore load balancing is not only carried out in a stretch of livelihood but also in a way that can be expressed in the formal manner as it is accepted by the occurence of infrastructure.

      The splitting of workloads in any transactions is suitable for execution and can be able to maintain the efficiency and responsiveness with concerned to the cloud infrastructure. Dividing the tasks to two or more servers increases the efficiency and responsiveness. Underlying too many tasks with a single server may leads to the

      VM1

      Guest OS Hypervisor

      VM2

      Guest OS Hypervisor

      performance bottlenecks. Therefore the configuration should be maintained for the workloads so as to carry out the tasks more efficiently[5]. There should be periodic translations between any two services. The recovery time should be maintained in order to get the clear idea for the requested clients and the service providers. The availability of resources should be provided with the impact of different strategies so as to improve the cost of performance that guarantees the request of user and service provider. Thus the penalty of impact of undertaking the requirements produced due to the effect of generalizing the responsiveness and stability to perform the actions carried out by the cloud infrastructure. The optimizing

      Fig 4. Interaction between VM and OS

      5

      4

      3

      2

      VM

      Guest OS

      5

      4

      3

      2

      VM

      Guest OS

      1

      0

      1

      0

      task 1 task 2 task 3 task 4

      task 1 task 2 task 3 task 4

      Fig 5. Graph representing capability between VM and Guest OS

      characteristics involves the variation identified by the virtual machine to co-operate with the task done by the service provider. Some of the assumptions that are reliable to this context are:

      • When the system queue is full, it accomodates the arrival of job to be redirected.

      • Normalizing clouds are characterized by an availability an.

      • Normalized clouds are also characterized by a quality level qn (0 < qn 1) that determines the quality of service reached by a request.

      • The time required by a VM to accomplish the work is T=1/ . Therefore the execution time is given by Tn=1/(qn.) T).

    2. Multiplexing VM

      In multiplexing, the number of running VMs are greater than Q , i.e., 0 D# run L and each one of the physical machine can be loaded with more than one virtual machine. Let us assume that the scheduling algorithm in round robin fashion is able to balance the load among the Q physical

      machines, the maximum level ' m ' achieved by each physical machine is given by:

      m = D#run / Q (1)

      Let us consider the cardinality K , i.e., |K| = D#run . The VMs can be partitioned into 2 sets Km and Km-1 with K = Km U Km-1 that corresponds between the two multiplexing levels. The cardinality of such sets is given by:

      run

      run

      |K| = m . [ D# – (m – 1) . Q ] (2)

      | Km-1| = ( m – 1 ) . [ m . Q – D#run ] (3)

      In this way the execution of virtual machine is carried out in different strategies. The average execution rate of finding the appropriate results is the major advantage of this virtual machine.

      Fig 6.Cycle which shows Multiplexing

      The above multiplexing cycle shows the creation of the configuration and deploying it in a cyclic manner to increase the flexibility of the virtual machine.The stages that include are:

      • Creation of VM builder

      • Discovering the created VMs

      • Check In of the version control

      • Edit the configuration of VM image testing

      • Deploy the development production

      • Monitor the status performance availability

      • Manage life cycle, cloning properties, location

      • Finally destroy the errors.

    3. Threshold Evaluation

    The threshold evaluation is the frequency which produces the responses and effects. It also involves minimizing the cost of paging translations in virtual environment. Some of the privileged instructions are trapped through the hardware mechanisms. CPU utilization may get vary by

    increase or decrease in the threshold frequency of the workloads that have been carried out in the frequent instantiation. The shortage of CPU and memory resources is done by the large amount of migration process[6]. Resource utilization keeps on changing with the accommodation of cloud services that are carried out between the physical machines and the virtual machines. Dynamic management of virtual machine plays a major role in frequency scaling. The physical machine should be swithed off in order to reduce the power consumption whereas virtual machine should be migrated from physical machines to undergo low resource utilizations.

    E = h (4)

    where E is the enery, h is the plancks constant and is the threshold frequency which is involved in photoelectric effect. The frequency of virtual machine also depends on the virtual machine monitor. The latest technology VMware player also includes in this as it is applicable for the operating system such as Windows 7, Ubuntu , etc. The better efficiency is based on the fact that the threshold frequency applies to the virtual memory. The foundation of windows azure platform also applies to the virtual machine manager.

  5. EXPERIMENTAL RESULTS

    In this section, some models have been proposed to quantify and illustrate the effects of different cloud-based strategies under different load conditions. Our aim is to provide a better infrastructure service to the virtual machines with the benefit of utilizing performance. In order to determine the performance some of the parameters are required by taking system elasticity into account. Let us define the system capacity U as the number of users that are concurrently managing the system. There are some logical resources K of the system which are further partitioned into B physical resources and C virtual resources such that K = B + C. Therefore the capacity of the system can be illustrated as follows:

    U = B + C + X + Y (5)

    Let us analyze a cloud system characterized by some of the initial parameters like B = 200 , C = 0 , X = 10 , Y = 0 ; therefore the result of the capacity of the system is given by U = 210 without multiplexing of VM. The queuing of resources should not be ignored once the task is completed since the resource will be finite state and it can be activated in any of the ready state. Though the analysis of resources can be ignored by the traditional performance but it cannot be ignored by availability, failures and recovery. The analysis of some resources such as availability, reliability, scalability, flexibility, scalability, extensibility is given in the table between the year 2008 – 2012. It can give users more control and the flexibility of managing their own systems, while providing the consumption benefits of cloud computing. Cloud computing originated as a new way to deliver IT services by providing a customer interface to

    automated, self-service catalogs of standard services, and by using autoscaling to respond to increasing or decreasing user demand. " A private cloud, in its own virtualized environment, gives users the best of both worlds. Dividing the tasks to two or more servers increases the efficiency and responsiveness.

    Table 1: Performance Results

    hardware layer, said John Livesay, vice president of InfraNet, a network infrastructure services provider. "This is done by installing a Hypervisor on top of the hardware layer, where the systems are then installed." "Virtualization is a foundational element of cloud computing and helps deliver on the value of cloud computing," Adams said. "Cloud computing is the delivery of shared computing resources, software or data as a service and on-demand through the Internet." Most of the confusion occurs because

    Category Year

    Availability Extensibility Reliabilty Scalability Efficiency

    87

    64 56

    37 32 91

    88 67 35 64 43

    Availability Extensibility Reliabilty Scalability Efficiency

    87

    64 56

    37 32 91

    88 67 35 64 43

    2012

    Year

    2011

    Year

    2010

    Year

    2009

    Year

    2008

    virtualization and cloud computing work together to provide different types of services, as is the case with private clouds. The cloud can, and most often does, include virtualization products to deliver the compute service, said Rick Philips, vice president of compute solutions at IT firm Weidenhammer. "The difference is that a true cloud provides self-service capability, elasticity, automated management, scalability and pay-as you go

    1. Quality based service

      The number of attempts are made by the service provider and the researchers to balance the cost of maintaining service performance and the profit of increasing resource utilization. The correlation between the service provider and the clients is identified by the resource utilization and performance characteristics. In a nutshell, virtualization is software that separates physical infrastructures to create various dedicated resources. It is the fundamental technology that powers cloud computing. "Virtualization software makes it possible to run multiple operating systems and multiple applications on the same server at the same time," said Mike Adams, director of product marketing at VMware, a pioneer in virtualization and cloud software and services. "It enables businesses to reduce IT costs while increasing the efficiency, utilization and flexibility of their existing computer hardware." The technology behind virtualization is known as a virtual machine monitor (VMM) or virtual manager, which separates compute environments from the actual physical infrastructure.

      Fig 7. Represents the Infrastructure service of VM

      Virtualization makes servers, workstations, storage and other systems independent of the physical

      service that is not inherent in virtualization."

      "Private cloud computing means the client owns or leases the hardware and software that provides the consumption model," Livesay said. With public cloud computing, users pay for resources based on usage[7]. "You pay for resources as you go, as you consume them, from a [vendor] that is providing such resources to multiple clients, often in a co-tenant scenario." A private cloud, in its own virtualized environment, gives users the best of both worlds. It can give users more control and the flexibility of managing their own systems, while providing the consumption benefits of cloud computing, Livesay said. On the other hand, a public cloud is an environment open to many users, built to serve multi-tenanted requirements, Philips said. "There are some risks associated here," he said, such as having bad neighbors and potential latency in performance.

      In contrast, with virtualization, companies can maintain and secure their own "castle," Philips said. This "castle" provides the following benefits:

      • Maximize resources Virtualization can reduce the number of physical systems you need to acquire, and you can get more value out of the servers. Most traditionally built systems are underutilized. Virtualization allows maximum use of the hardware investment.

      • Multiple systems With virtualization, you can also run multiple types of applications and even run different operating systems for those applications on the same physical hardware.

      • IT budget integration When you use virtualization, management, administration and all the attendant requirements of managing your own infrastructure remain a direct cost of your IT operation

        While virtualization is the best solution for some organizations, a cloud solution offers several benefits that are more suitable for other businesses.

      • Outsourced IT The day-to-day administration, care and feeding of supporting systems move away from you to the service provider. This could free up internal IT resources for higher-value

    business support and allow you to put IT budget dollars toward efforts that advance your business.

    • Quick setup Cloud startup is relatively quick and easy. Plus, servers, appliances and software perpetual licenses go away hen you use such a service.

    • Pay-as-you-go An example could be found in Software-as-a-Service (SaaS) applications available today that allow the off-loading of basic IT requirements to cloud service providers. You pay for what you need and use. But you do not have to continue to invest in many of the products used to support the network and systems, such as spam/anti-virus, encryption, data archiving, email services and off-site storage.

    • Scalability By using the cloud, you can also temporarily scale your IT capacity by off-loading high-demand compute requirements to an outside provider. As a result, as mentioned above, you pay for only what you need and use, only at the time when you need it.

    Many companies are already virtualizing their IT environment and have been doing so for years. Initially, virtualization was deployed for compute resources, primarily as a cost-saving technology. Organizations soon recognized that virtualization provided additional cost- savings benefits as well as enhanced speed and flexibility. Most clouds are built on virtualized infrastructure technology. Cloud computing originated as a new way to deliver IT services by providing a customer interface to automated, self-service catalogs of standard services, and by using autoscaling to respond to increasing or decreasing user demand. From an IT perspective, a private cloud offers the key advantages of speed, agility, and efficiency while maintaining control of sensitive workloads.

  6. CONCLUSION AND FUTURE WORK

This paper enhances to provide requested services to the clients through cloud infrastructure by maintaining performance in VM. First and foremost thing to look after is that performance should be stable in VM if it varies then it might be a problem to provide service through the cloud. Most of the confusion occurs because virtualization and cloud computing work together to provide different types of services, as is the case with private clouds. The cloud can, and most often does, include virtualization products to deliver the compute service. cloud provides self-service capability, elasticity, automated management, scalability and pay-as you go service that is not inherent in virtualization."

Future work can be done on other cloud services such as Platform-as-a-service and Software-as-a-service to provide services based on the impact of performance stability in VM. Already research has been carried out by PaaS and SaaS in terms of stack solution . So work can be done on these by looking into Capacity , Elasticity , Availability and many other resources that are included in virtual machines. Not only virtual machine physical machines must also be considered into account. Therefore

the final result must be better utilized for the usage characteristics of the cloud service to the clients.

REFERENCES

[1] H. Khazaei, J. Misic, and V. Misic, Performance analysis of cloud computing centers using m/g/m/m+r queuing systems, Parallel and Distributed Systems, IEEE Transactions on, vol. 23, no. 5, pp. 936 943, may 2012.

[2]A.V.Doetal.Profiling applications for virtual machine placement in clouds, in Cloud Computing (CLOUD), 2011 IEEE International Conference on, july 2011, pp. 660 667.

[3]A. Iosup, N. Yigitbasi, and D. Epema, On the performance variability of production cloud services, in Cluster, Cloud and Grid Computing (CCGrid), 2011 11th IEEE/ACM International Symposium on, may 2011, pp. 104 113.

  1. M. Mishra and A. Sahoo, On theory of vm placement: Anomalies in existing methodologies and their mitigation using a novel vector based approach, in Cloud Computing (CLOUD), 2011 IEEE International Conference on, july 2011, pp. 275 282.

  2. B. Rochwerger et al., Reservoir – when one cloud is not enough, Computer, vol. 44, no. 3, pp. 44 51, march 2011.

[6]H. Liu et al., Live virtual machine migration via asynchronous replication and state synchronization, Parallel and Distributed Systems, IEEE Transactions on, vol. 22, no. 12, pp. 1986 1999,

dec. 2011.

[7]R. Ghosh, K. Trivedi, V. Naik, and D. S. Kim, End-to-end performability analysis for infrastructure-as-a-service cloud: An interacting stochastic models approach, in Dependable Computing (PRDC), 2010 IEEE 16th Pacific Rim International Symposium on, dec. 2010, pp. 125 132.

[8]R. Ghosh, F. Longo, V. Naik, and K. Trivedi, Quantifying resiliency of iaas cloud, in Reliable Distributed Systems, 29th IEEE Symposium on, 2010, pp. 343 347.

[9]S.Ostermann et al., A Performance Analysis of EC2 Cloud Computing Services for Scientific Computing, in Cloud Computing, ser. Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering. Springer Berlin Heidelberg, 2010, vol. 34, ch. 9, pp. 115131.

[10] X. Meng et al., Efficient resource provisioning in compute clouds via vm multiplexing, in Proceedings of the 7th international conference on Autonomic computing, ser. ICAC 10. New York, NY, USA: ACM, 2010, pp. 1120.

[11]R. Buyya, R. Ranjan, and R. Calheiros, Modeling and simulation of scalable cloud computing environments and the cloudsim toolkit: Challenges and opportunities, in High Performance Computing Simulation, 2009. HPCS 09. International Conference on, june 2009, pp. 1 11.

Leave a Reply

Your email address will not be published. Required fields are marked *