Snapshots to backup files - this is my plan, does it make sense?

Hi,

digging through the forum, having posted a question about snapshots and having researched some, I still struggle in understanding how snapshots truly work.

I would like to write down the plan I have in mind for snapshots and how I would configure them and then I would like to ask you guys to comment on my plan!

Real scenario, currently installed: I have a storage capacity of about 1.8 TB. I have allocated a target of 1 TB. I have allocated 700 GB for snapshots, 100 GB free for maybe setting up another target as a swap drive or something. My iSCSI storage server is connected to my host system with 2x GBit (802.3ad).

On my 1 TB target, I have numerous containers of virtual machines (VMware). Right now I have 3, with one more coming in two weeks any maybe another 2-3 over the next few months. These containers take about 200 GB in total right now, growing as the time goes by. All VMs are totally isolated inside their containers, they do not have any shared data pool or anything via the target volume, they use normal Windows shares for the little data they share.

My backup plan is the following: Once a week, the host system the VM machines run on (booted on its own HDs - not an issue since if I ever plan on booting it off the iSCSI, I will set up a dedicated target for that I guess) will backup the contents of the entire iSCSI target ( = the 200 GB) to tape. So I have weekly backups, using different tape sets I have several weeks of backups effortlessly.

Now for the daily backups I want to use the snapshot functionality of my iSCSI server.

I set it up like this:

- 6 Snapshot sets (on day 7 I have the tape backup)

- each snapshot is set to be running from 7 AM to 11 PM ( business hours + late working drones)

- each snapshot set has 1/6th of the 100% of my 800 GB snapshot space = 16,7%

Scenario: If I ever discover some user has messed up/deleted a file on monday and it's thursday:

I mount the snapshot target of sunday, copy the entire VM file of the concerned server to a separate disk, start it in an isolated environment, recover the file in question, replace it on my running server, and then discard of my local copy of the file again.

If I am completely wrong, please elaborate on where my mistake is and please provide me with a hint on how to establish the backup plan I have with my ressources!

I'm really challenged in understanding what the timeframe of the snapshots is good for, so I put it to use the best way I could think of.

Thanks for being patient and answering my endless questions, I'm just figuring all of this out! :-)

You will need to have separate running Snapshots for each task or day. If Snapshot is scheduled for the day that he deletes files then the Snapshot will perform its process for that volume and if the data is deleted then Snapshot will overwrite its existing data with the current and you will not be able to restore as the Snapshot will record the volume with the deleted data. You could remove the Snapshot from the Target Manager and when you need to restore just enable again, just as long as the Snapshot is not scheduled to run during the time the data was deleted.

Todd,

thanks for your efforts to explain snapshots to me. I'll just have to play around with them to understand them to a level where I can implement a fool proof backup strategy.