• State Suggested Answer
  • Replies 6 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 6310 views
  • Users 0 members are here
Options
Related

Live restore large volume to new OS on different server?

msmith_81554
over 5 years ago

Here's the scenario, I want to see if I'm thinking straight here:

I have a file server with a very large data partition of almost 5 TB.  It's running Windows Storage Server 2008 and getting low on physical space so I want to move the data to something more modern, specifically Server 2012 R2. The server name needs to stay the same as the old one because our document management software depends on UNC paths and it's not easy to change.

My plan is to create a new 2012 VM on different hardware, turn off the existing 2008 server, name the 2012 server the same as the old 2008 one and assign the same IP, and install the RR agent.  Then I'd kick off a Live Recovery for the data partition, thereby keeping downtime to a minimum so users can access the data even while it takes days to restore.  Is there any reason that would not work?

Assuming that plan works, what happens after the restore? Does RR take a full base image of this huge server, or is it smart enough to know only the C: drive changed and the D: drive is just an incremental backup?

Thank you in advance for any thoughts and advice.

  • 0 over 5 years ago

    Another option would be to use robocopy (or something similar) with the /r option (I think) this allows you to run the copy several times and it skips what is can not get and only transfers what has not been done already. For example 

    robocopy /r #1 runs for 24 hours and gets 80% (it skips files that are in use)

    robocoopy /r #2 tries to copy any new files and the 20% it missed in the first run. Lets say it gets 10%)

    now you stop access to the main server (so files are not in use) and run robocopy #3

    You don't have issues with deduplication, or middle men (RR) and you don't have to worry about disk size mismatches or other issues. Run robocopy as many times as you want till you get the smallest amount of data left

  • 0 over 5 years ago

    I think a live restore would work like that but "seeding" the volume with robocopy like Emte suggested is pretty common practice. I live restored a heavy use server in production and no one every was the wiser. It does copy everything from the recovery point regardless of what is still there in the volume you are recovering to. I had hoped it would only do changed data but it does it all.

    I think you will get a new snapshot for sure with new hardware regardless of how much data changed.

  • 0 over 5 years ago

    As long the data partition was on a different volume than the system volume, you can do a live recovery of just a single volume. You will need to have the 'new' target volume be the same size or larger than the one that it had been on (with a VM even if you won't use the whole thing, just make it larger and thin provision it). Also for live recovery you must use an agent (you mentioned that, but just to clarify you will need an agent). The other 2 suggestion (Emte and Corrigun) are spot on, but you do have options and RR is one of those should you choose to use it. Works just fine, matter of fact that almost the text book idea behind the Live Recovery option. 

  • 0 over 5 years ago

    Hello,

    I did something similar and it worked well, We had an old 03 r2 file server that had a file share volume and I restored the volume to a new different server 2012 r2 . I setup a new server with a c volume and another volume for the file share, installed the agent. When the time came for switch over I think we renamed the old and changed its ip and renamed the new and changed its ip and then started the restore.

  • 0 over 5 years ago in reply to ericw

    it was like 450gb worth of data, I think the job failed but I restarted it and it finished the second time

  • 0 over 4 years ago

    I know this thread is old but I want to do something similar but haven't decided the best way to do it. Old server has a 7TB partition and want to move it to a new server, but my servers will have different names. I was thinking of mounting the latest backup, using robocopy to seed the new server and maybe using DFS to replicate any changes. Since the 7TB partition contains mainly very small files, I am worried that the robocopy will take forever, especially if it runs into errors (path too long, etc). I am wondering it restoring the latest backup to a new volume on the new server would be quicker than using robocopy.