I think you're mixing two concepts together. I know virtual machine concept is an integral part of Cloud Computing but "Virtualization is a separate concept as well" and you can treat it separately. So can you please let me know you're asking parameter if Virtual Machines in Cloud Computing or the parameters in Virtualization? I think you're working on Virtualization in cloud computing. Am I right? Please clarify and I'll try to answer your question.

Resource (i.e. processor, memory also some kind of specific peripherals), Platform (i.e. OS) and Configurations (e.g. such as installed libraries)

It mainly depends on the user application requirements. For example, Amazon EC2 offers different instance with different type; each type is suitable for a certain application. Other services such as HPC requires require high bandwidth, low latency networking, and very high compute capabilities. Importantly, high I/O performance instance are for applications that are considered as communication intensive such those programed using MPI model. MapReduce application is a data -intensive application and requires high memory, and low network latency.

In short, Knowing the requirement of your application is the key to ask for a particular environment from a provider.

The first question you've asked is about the specification of the virtual machine (VM). It is about what the provider is offering to the consumer (buyer). I assume you are referring to the virtual machine being offered by cloud providers such as Amazon EC2, GoGrid, Google Compute Cloud. The parameters in this case typically include cpu, memory, storage and generally charged by per-usage (e.g per-hr) and also the bandwidth pricing. Some include disk transaction price. Of course, it includes operating system such as Windows or Linux, and some are bundled with other application software such as MS SQL or Ubuntu with Tomcat.

However, if you are referring to the virtualization technology, such as VMWare, XEN, that would be a different case.

The second question you've asked is about the user needs (application's requirement). That would varies according to application. For example, if you are doing a media streaming application, other than standard OS requirement, you might be looking for streaming server software or any other transcoding capabiities. You might want to read the following papers:

http://www.chinacloud.cn/upload/2012-11/12110409165965.pdf

https://www.researchgate.net/publication/230765183_Mapping_Application_Requirements_to_Cloud_Resources

Conference Paper Mapping Application Requirements to Cloud Resources

thanks to all of you

I want to clarify my question...

if we want to create some virtual machines on a physical machine what parameters (from physical machine) are required.

Hello Brijendra,

Your question needs to be answered in two parts. The first scenario is if you are designing and deploying a virtual infrastructure (VI) within your organization. For that you will have to consider a number of factors:

1. Networks

2. Storage

3. Converting physical servers to virtual

4. Size virtual machines based on application requirements (cpu, memory, storage)

5. Vendor and management tools (VMware leads in this area)

6. Include design considerations to allow for growth, disaster recovery, resiliency and availability, software currency

7. Consider backup options

8. Take into account cost of licensing for both virtual infrastructure provider and OS licences

9. Training of IT support staff

10. Co-locate servers or host internally

However, for the second scenario where you would like to buy cloud computing from an established service provider like Amazon, Rackspace, Google etc rather than setup your own VI.

For such services you may have to consider the following:

1. Vendor

2. Cost and billing

3. Availability of tools for the platform

4. Security

5. Initial sizing of virtual machines

6. Future increases in load and traffic are handled transparently by elastic nature of the cloud

As you can tell, the second option transfers most of the design and support issues to the service provider.

Hope this helps.

Asaf

Hi Brijendra,

Regarding your recent post "if we want to create some virtual machines on a physical machine what parameters (from physical machine) are required."

If I understand correctly, you want to know the specifications of server that will host the virtual machines. There is no correct answer but does depend on the number of virtual machines you plan to run. Let me try and answer this.

1. A server system with multiple CPUs and cores (2CPU/4 core would be a start)

2. Typically 64GB or more of RAM

3. At least 4 NICs (8 or more will allow redundancy)

4. Choice of standalone or blade system from IBM/HP/Dell/Cisco etc

5. Hardware to support connectivity to storage (iSCSI/FC/NFS)

6. The host CPU (AMD/Intel) should support virtualization

7. The BIOS and firmware should be current in order to run the latest version of virtualization software

8. The host machine allows booting from USB or local disk or SAN

As a start you will reference a compatibility matrix provided by virtualization vendor to identify which hardware is certified to run on a particular version.

Hope this helps.

Asaf

Hi Asaf Maruf,

thanks for your response....

suppose we have all type of support and we have to only devide the resource among VMs, what will be the resources which we have to take care?

Ex. memory, cpu etc.

Hi Brijendra,

If I understand you correctly you want to find out what resources are shared between VMs. The VMs practically share all resources of the virtual infrastructure including virtual switch, host NICs, network, datastore, memory and CPU.

The vSphere scheduler transparently manages CPU and memory resources among competing VMs. The VM administrator can establish resource pools and shares to ensure resource requirements of VMs are met and the environment runs smoothly.

Hope this answers your question.

Asaf

According to my experience it is important to have enough memory on the host computer. VMware can be installed and run on many host OS : XP, Win 7, Linux, SUZE Linux, OpenSuse 11.1, Ubuntu 9.04. I recommend test you VMware v.6.5.2 Workstation by VMware, Inc. and also VirtualBox v.3.0.4. And, don't forget to install drivers for all peripherals which you used on the virtual machine also: printer, scanner, ... These drivers must be installed under guest OS although were installed on the host machine.

Thank you Mr. Asaf Maruf

yes, you are right.

Hi all,

I have one more question....

I want to simulate resource allocation problem in VMs. can anyone please suggest any simulator?

Thanks

Hi Brijendra,

I don't know of a simulator that would be able to provide the simulation you require, hopefully one of the researchers will be able to answer.

That said, one of the advantages with virtualization is the ability to quickly setup a lab environment with minimal cost and resources. You should be able to "simulate" resource allocation problem in your virtual environment by assigning resource pools, memory and disk shares to virtual machines and subjecting the virtual infrastructure to network and I/O traffic.

Best of luck.

Asaf

thank you Asaf...

Hi Brijendra Kumar,

Try CloudSim for simulation http://www.cloudbus.org/cloudsim/

Well, it depends what exactly you need to do during your experiments. Though there are general metrics and parameters as you mentioned such as number of processing entities (PI), CPU speed, memory size, bandwidth size, and so on which are already defined in simulators such as CloudSim, but you can add further more parameters to enhance your research outcome. CloudSim is a java based simulator and you can add whatever you like to it to fetch required results. Hope it works.

For your second question of how to test and analyse the resource allocation algorithm, I guess this is what you need to simulate right? You have to code your algorithm in CloudSim and see what is the results of the algorithm in terms of utilization rate, speed and latency, and so on...

The very important parameter is the hypervisor which is the important parameter to control all the task in the vms and most of the time it is also called as middleware which is very important for scheduling

At Kiruba Karan: Hypervisor is not a parameter. It is Virtual Machine Manager (VMM) which plays the role of managing all VMs in a virtual environment. Parameters are defined for each VM and Hypervisor is responsible to allocate those parameters to each virtual machine. Isn't it?

Saeid Abolfazli: yes, that's correct...