Ideal Candidate for P2V

[jitendrakmr]

Hello Everyone,

I am working on a P2V project under which I have to migrate approx 50 Physical servers.I had configured perfmon monitors on physical servers to identify right candidates for virtualization. I analyzed perfmon logs and in doubt following servers -

1. ServerA - " % Processor Time" counter data of 7 Days is as follows -

Min	Avg	Max
78	83	100

This physical server has dual Intel processor @ 930MHz. As can be seen that server has high CPU utilization but server is quite old.Please let me know if it is a good idea to virtualize it to ESX host with 4x2.4GHzAMD cpu?

2. ServerB - While analyzing logs I noticed that average PTE (for 7 days) is approx 6000 in this server however available memory counter seems okay and also the server has average network usage (in and out) approx 5MBps. Please let me know if low PTE counts make any difference for P2v? and is 5Mbps avg network usage okay to be virtualized?

Thanks in advance!

Jitendra Kumar

VCP, MCSE 2003, MCITP Enterprise Admin, CCNA, ITIL Foundation, Netapp NS0-153 (Storage Networking)

Personal Website - http://www.virtualizationtrainings.com, http://www.hillsindia.com

[kjb007]

1. Server A, with dual 930 MHz, is effectively using 1.86 GHz, which is less than one core of your new processor. I think you will find this vm to perform better than it did previously, as I have with some of the older servers I virtualized. Your mileage will vary of course.

2. Not sure what you mean here with ServerB. None of the numbers appear very high. If it doesn't work well, you can always go back to physical.

-KjB

[suprauche]

Hello,

In your situation i would not put ServerA in an ESX server. It has a high CPU utilization rate for a virtual environment. It sometimes take 100% CPU means it can go further and could dominate your ESX server.

The second server seems to be ok. 5 mpbs is a very low value if you use a gbit interface.

In my opinion the best candidates for an ESX server is test and development servers which uses lots of physical machines and the needs grow day by day. Also the passive nodes of an active/passive cluster could be a good candidate for a disaster recovery plan.

Best Regards,

[weinstein5]

I agree with KjB's assessment - for Server A yes it is a high utilization but it is an old processor with limited cycles - so moving it to brand new hardware you will see a perofrmance boost - and should be able to run it on a single proc - Server B also does not look loaded and should be fine as well -

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful

[kjb007]

I would disagree with you. In some cases, test and dev servers, where lots of code compilation and several applications and users actively developing apps can be heavier users of resources than production machines, which typically have one function.

I runs hundreds of production servers as virtual machines, and have recovered unused physical assets and have made the environment more efficient, and built-in high availability and DR capabilities as a default configuration.

-KjB

[suprauche]

It depends on the characteristics of your test server. If the server is being used for performance test purposes, of course it is not a good candidate. But if it is being used for function tests and does not take CPU all day, why would not i use that. Also usually there is no activity on test platforms at night. So within DRS, you can use your ESX servers more efficiently in that case.

[kjb007]

The point of the post that a blanket statement to describe candidates does not always work. You can't say test&dev make the best candidates, nor can you say the prod does NOT make good candidates. The true answer is, that it depends. You pointed out test&dev, where I pointed out a scenario where that does not fit. It all depends.

-KjB

[suprauche]

I agree with you KjB. It all depends. But when i have to think about a p2v conversion, i first think about if it fits the general rule of thumbs. And i use a matrix probably you know from web. It does not always tell the truth but it can be useful if you can not decide in some cases. I would like to share it with you.

Regards,

[jitendrakmr]

Thanks a lot guys.

I will give a try to virtualize ServerA soon and will post here my findings. I also have some linux box to virtualize. Does anyone has any experience on P2Ving Linux box using Vmware converter (cold migration)? If yes then can you guys please share your experience and possible challenges?

Thanks again.

Jitendra Kumar

VCP, MCSE 2003, MCITP Enterprise Admin, CCNA, ITIL Foundation, Netapp NS0-153 (Storage Networking)

Personal Website -

[suprauche]

You know you can not convert the linux guest while they are running. So what i do is, i take the image of the linux server with acronis and using vmware convertor i convert it to a virtual machine. It is a very simple process. But you must consider something here. During this conversion process i could not resize the linux filesystem for the destination host. For example you have a 40 gb partition on your linux box. Than you will have a 40 gb disk capacity in your guest system. I could not used to pass that problem. Maybe the new version of the convertor or another method could bypass that problem. I do not face that problem during Microsoft systems. I can resize the partitions without any problem.

Regards,

[kjb007]

I typically don't P2V linux vm's, and build them from scratch using a template. The process posted by suprauche will work well also. The converter 4 beta has support for Linux conversion, so you should search for it and download that to try out.

-KjB

[jitendrakmr]

YesI understand that cold cloning is only option for Linux P2V migrations and also understand that volumes cannot be resized in case of linux guests using Vmware converter.

What I am looking here is that 1. once we boot the physical server using Vmware converter CD then does the P2V process completes easily without giving any trouble or does it work only for specific versions of linux? if it works just for some specific versions of linux then can I get that list somewhere? 2. Once we boot the VM created after P2V does linux server do post migration hardware/device driver configurations automatically? or do we need to do it manually?

Thanks.

Jitendra

VCP, MCSE 2003, MCITP Enterprise Admin, CCNA, ITIL Foundation, Netapp NS0-153 (Storage Networking)

Personal Website -

[kjb007]

The cold-boot cd will get your information copied over. But, since it's cold-boot, no driver configuration is done. So, if all goes perfectly, your vm will start up, and kudzu will run, an all of your devices will be recognized, and hardware updated. This is not always the case, and there's no supported guest section for Linux. So, keep your rescue cd handy, as you may need to go in and fix drivers, especially your scsi controller.

-KjB

[Kahonu]

What do the numbers on the X-axis represent??