locked
Do I need to backup my VM? RRS feed

  • Question

  • I just cannot get my head around this in the land of the cloud. Any cloud.

    My main reason to look at cloud VM (at the moment) is to take hardware off my worry list.

    I have a server in a DC for a small business and would love to take an image of it and create a VM on the cloud. (how details not important for now)

    So...

    Does the cloud provider in general or say MS with Azure specifically, provide the "don't worry if the hardware fails, you won't notice a thing" scenario? Or do I still have to set up VM level backups. I mean a VM is essentially one big file - right?

    By "VM level backups" I do mean the highest level backup and not data only backups within the VMs OS. I would of course implement "normal" user data backup strategies.

    I ask this as when I was playing with the Rackspace cloud VM the intro talked vaguely about "if you are migrating from a physical server then you also need to...backup the system...etc" I may have misunderstood but it sure looked like I needed to take into account the possibility of their hardware failure by having a VM backup strategy in place.

    I you haven't fallen asleep...

    So any thoughts appreciated.

    Cheers

    Tom

    Monday, June 17, 2013 7:36 AM

Answers

All replies

  • Hi,

    Azure Virtual Machines use blob storage for both OS and data disks so we can get some basic backups going with nothing more than the blob API and a simple powershell script.

    http://blog.greatrexpectations.com/2013/04/24/using-blob-snapshots-to-backup-azure-virtual-machines/
    Friday, June 21, 2013 9:42 AM
  • Hi Tom,

    Great Q.

    Short version : IaaS is a bit dangerous cos theres no tape / Storage guy giving you a hard time when u build a machine to make sure there is a recovery point.

    A friend of mine* , who was fairly experienced in Azure / Virtual Machines etc recently deleted the wrong Virtual Machine and all the storage files by mistake using a really cool Powershell script they wrote... It was a throw away , test , Proof of concept type VM so it didnt matter.

    However, got me thinking bout what be the recovery plan / Recovery Point if it was real life ?

    Would you phone Microsoft Azure support begging for backups of the VHD ?

    Of course you would just build a new image from the Gallery, re-install all the software , registry settings , congfig and all would be ok , cos since the server was built , the process to do this would be fully documented / scripted , taking into account and software releases, patches, config changes - yeah right.

    I guess the problem is that it is hard to back VHD's out of the cloud etc cos they are so big and to do so sort of defeats the object in some way.

    So mitigation is

    - Good Change Control / Tracking. Ie you should have a documented, perferably scripted process to rebuild a server to the current level

    - With Proprietery / complicated / lots of config type VM's consider Imaging them into Azure periodically (and perhaps offsite as an Image)

    and testing above works

    I think i saw something recently about delayed VM replicas for on Premise Hyper-V , ie you are constantly shipping changes offsite , to a different datacentre, but changes either aren't applied , or are snapshotted to a point in time so you can recover from screw ups. That would be cool in Azure.

    Sorry not an answer !

    cheers

    Andy

    Thursday, June 27, 2013 7:09 PM