VMware Cloud Community
Shoganator
Enthusiast
Enthusiast

Choices? Virtualise SQL servers which use SAN LUNs P2V and RDMs

Hi all,

I have never had the situation where I need to virtualise a system which has fairly large data disks which are presented via SAN LUNs. I am hoping to get some views/ideas as to the process to follow. I guess I have a couple of options here:

Currently, the largest data disk in use on the system in question is 1073GB in size. For a vSphere 4.1 environment, this would mean it is comfortably within the 2TB minus 512 bytes maximum VMFS volume size (considering an 8MB block size formatting). The other data disks which are also presented via SAN LUNs are smaller in size and could fit on other datastores once P2V'd. Based on this info, I would say that it would be preferable to P2V everything and not worry about using RDM.

I would like to present the option for RDM though, seeing as though the data disks are already being presented from LUN, or if not, at least get a good idea of the process for consolidation in this type of scenario.

I have done my research and gathered data on performance metrics - CPU, Mem, network and disk. All are within scope for virtualisation - disk usage is fairly high though, with the average IOPS utilisation over a 24 hour period being 395 IOPS - (peak of 490 IOPS).

What are your thoughts on the above two options? Which would you go for, and would someone be able to give me an idea on what they would do should they want to virtualise this system and use RDM for the data disks? - what would the high level process be?

Thanks for any input Smiley Happy

My personal blog: http://www.shogan.co.uk .::. Twitter: shogan85 .::. if an answer has helped or solved your question, please don't forget to mark as "Answered" or "Helpful"!
Reply
0 Kudos
2 Replies
vmroyale
Immortal
Immortal

I tend to convert the OS volume and then present the RDMs back to the new VM in these cases, but it usually due to time constraints for the conversion.

The current setup you have leaves ~50% room for growth, and with vSphere 4 the 2TB limit is going to be there regardless of the RDM or VMFS approach. I would think about how you want to deal with the data growth and what to do IF/when the data exceeds 2TB minus 512 bytes. With vSphere 5, you can use physical mode RDMs up to ~64TB so this solution can scale where VMDKs are still limited to 2TB minus 512 bytes.

The biggest thing (IMO) is to think about how backups are going to work. Physical mode RDMs can't utilize snapshots, so that is one tradeoff for having larger volumes.

Brian Atkinson | vExpert | VMTN Moderator | Author of "VCP5-DCV VMware Certified Professional-Data Center Virtualization on vSphere 5.5 Study Guide: VCP-550" | @vmroyale | http://vmroyale.com
Shoganator
Enthusiast
Enthusiast

Brian,

Great post - thanks for that. Just what I was looking for. I was thinking along similiar lines with regard to presenting the RDMs back to the new VM - as it would be a service down and there would be time constraints with regard to how long it can be down for.

It's really helpful to discuss things like this as you always tend to get discussions about the pros and cons - like your example of backups and snapshots on physical mode RDMs Smiley Happy

Thanks again for your input, much appreciated.

Cheers,

Sean

My personal blog: http://www.shogan.co.uk .::. Twitter: shogan85 .::. if an answer has helped or solved your question, please don't forget to mark as "Answered" or "Helpful"!
Reply
0 Kudos