Disaster Recovery w/ Virtualization Setup?

[pctechit]

I am now thouroughly confused after spending all day today researching so many threads here. I am thinking in circles and need to talk it out.

Client: Small Business (5-35 nodes) Windows Only

Goal: I am looking to setup a disaster recovery plan that will work on most any Windows network. The criteria are as follows:

1.) Take a nightly image (complete) of all the workstations
2.) Ability to quickly setup that image on a virtual machine if needed
a.) if this could be automated great, or at least done remotely?
3.) Take that image and install it on new hardware if need be.

What do I need to purchase for each setup?

I cannot get my head around the entire concept. Between Acronis True Image, Shadow Protect, Clonezilla, Macrium Reflect, DriveImage XML, MS Home Server... who knows what else.

Then there is the whole Bare Metal issue and getting something that I can easily swing the image over to run a temporary Virtual machine while I work on the failed system...ahhhh!

Data Overload... any ideas?

 

[DarDar]

Well clonezilla has a server option that would allow for this. Just you need the pxe option available on machines if you want to do it remotely to each machine. It will back up to a shared drive and you could load an image. You could easily set a virtual machine to boot from an iso of clonezilla and then point it to the img file to load that img into the virtual machine.

The latest version of acronis has an option to turn the .tib file into a virtual image that is suitable for MS Virtual PC or Sun Virtual Box.

There is another open source tool that will image over the network but i can't remember the name right now.

Having said all this, why do you want a nightly image of the workstations? Would it not be easier to store info on a network share? You might also want to consider some form of offsite storage for these images. If anything were to happen at the site and you lost everything it would all be in vain.

 

[pctechit]

Most of our clients are using Folder Redirection and Roaming Profiles so I guess a nightly image is overkill. I was just trying to be overly cautious.

I like the Clonezilla route as I am trying to get this done as inexpensive as possible.

Any more resources on setting up the machines to PXE boot and dump the clonezilla image?

 

[basic]

I also agree that a nightly image of every single workstation is overkill and will be difficult for you to manage, especially if you are going to archive past images. I would just create a baseline image of the software baseline and set it up so that data is saved to the server and just do a nightly image (incremental, differential, full, whatever makes the most sense) of the server. I also hope that most if not all the workstation are the same model to make life easier for you.

 

[jaydeee]

have a friend addicted to acronis... its always his option for this kind of job but I also don't know it will fit to your whole system.

 

[sjc@blainecs]

2.) Ability to quickly setup that image on a virtual machine if needed
3.) Take that image and install it on new hardware if need be.

Then there is the whole Bare Metal issue and getting something that I can easily swing the image over to run a temporary Virtual machine while I work on the failed system...ahhhh!

I wanted to try and clear some up of the information above. This is general info and not specific to any application, so some may handle them different.

One of the big sticking points in Bare Metal type restores is hardware, specifically the HAL that is running on the machine. If the customer has Machine1 and a backup and you try to recover it to Machine2 with a different HAL the restore may complete but the server will not boot. There are some work around's to this but they are a pain and not very reliable.

Here is a TN about HAL's

http://support.microsoft.com/kb/309283

Most VM software does not have this issue as the VM simply sits on-top of the current OS. This is a great recovery method as long as you test your VM's and more importantly the process.

The process of backing up the OS is another issue. I agree that performing a full backup each night is a significant undertaking. But as far I as I know there is no way to build VM images from a INC type backup. Meaning if you do full backups of the OS for a VM image on Sunday and INC backups every other night. If you had a disaster you would drop the VM from Sunday on a computer and then have to apply the changes via an INC restore. Not terrible but also extra steps and prone to trouble in a full server restore.

If it was my servers and they were critical I would architect it so the network and more importantly the destination for backups was fast enough (many people try putting in a 2TB USB drive with a single or double spindle and that's never going to complete) to perform a full VM image each night. If the servers were not critical I would create an image once a month (or after any significant changes to the OS) and then perform CDP (Continuous Data Protection) on the user data. CDP performs backups all day long as a files change vs normal backups that wait till 8PM and then perform one backup.

Backups are a pain, and with the changes Windows is making in their products like Vista and Windows 7 is getting much worse for Enterprise's (home users its better)

I hope this helps and let me know if you have any questions.

sjc
Blaine Computer Services

 

[pctechit]

Thanks all, great info.

So to update what I am doing, right now I am working in my Virtualbox Lab. I have just setup my Ubuntu Server and installed DRBL (which has Clonezilla Server) tomorrow I am going to see if I cannot get an image from a system, then remove the Hard Drive from that system and PXE boot to the image on the DRBL server.

 

[DarDar]

Sounds like you are well under way - let us know how it works out!

 

[pctechit]

And just like that, the dream dies So I was able to get DRBL (Clonezilla Server) setup on my Ubuntu Server and was able to grab an image from a remote system via PXE boot. The "Diskless" portion of this setup will only boot the Ubuntu image though, so I am not able to boot into the Clonezilla image over the network.

Anyone have any ideas on what I can do with this clonezilla .ag file now? Virtual Machine? am I missing something with the network diskless boot side?

Also, the best tutorial I found for the DRBL server setup is here:

http://searchenterpriselinux.techtarget.com/tip/0,289483,sid39_gci1317237_mem1,00.html?ShortReg=1&mboxConv=searchEnterpriseLinux_RegActivate_Submit&

 

[rurbaniak]

I am now thouroughly confused after spending all day today researching so many threads here. I am thinking in circles and need to talk it out.

Client: Small Business (5-35 nodes) Windows Only

Goal: I am looking to setup a disaster recovery plan that will work on most any Windows network. The criteria are as follows:

1.) Take a nightly image (complete) of all the workstations
2.) Ability to quickly setup that image on a virtual machine if needed
a.) if this could be automated great, or at least done remotely?
3.) Take that image and install it on new hardware if need be.

Nightly images sound good, but I think it's unrealistic to do this for all workstations. First of all, if you need to, let the client know that a workstation should not be storing files, they should all go the server. Then recovering a workstation just for windows isn't that big of a deal, except for downtime.

Now, I understand when certain workstations is setup in a very specific way that doing a reinstall could be a pain. Having a backup of those on a periodic basis is the way to go. Do you really need to backup those daily?

I would forget about backing up everything daily. Just identify the critical workstations, and have a full backup of servers. Then have some images of basic workstations to speed up the process when a non critical workstation goes down.

Now, if a real disaster happens, like the place burning down, what would you do?

Your disaster plan should incorporate a plan to recover from all kinds of disasters, as long as the client is willing to do so. That would include, backups, offsite backups, backup hardware, and virtual images. I personally do this kind of disaster plan for several clients now, and if they burnt to the ground, we'd be able to take a backup server to ANY location with ANY type of computer for a workstation, and have them backup and running in hours.