Veeam Backup and Replication hardware

Because of the lack of characters in twitter, I thought I would share my setup via my blog and ask for recommendations on configuration.  Specifically the recommended RAID configuration for the best backup time.


  • VMware Essentials Plus, three ESXi 4.1 hosts
  • Identical Hosts:  Dell R710, Dual Intel Xeon E5530 2.4 GHz (dual quad core processors), 32GB of RAM
  • EqualLogic PS5000e iSCSI SAN.  16x 1TB SATA II 7200RPM hard drives in a RAID 50 with 10.47 TB usable
  • 23 Virtual Machines
  • Approx 7TB file server storage in use

Destination (Veeam Backup Server)

  • Physical Server: PowerEdge 2950, Dual Intel Xeon 5050 3.0 GHz, 4GB of RAM
  • Direct Attached Storage:  MD1000, 15x 2TB Western Digital RE4 Enterprise/RAID SATA II 7200RPM hard drives

So, basically a few questions:

  1. What RAID for best performance?  My research tells me RAID 10 has best performance.  However, I’d like to hear the recommended RAID setup for Veeam B&R.  Could be that RAID 50 is recommended and RAID 10 would be overkill given the way Veeam backs up?  I’m not sure.  I need a reasonable backup window. 
  2. Is the RAM okay, or need to upgrade?  This server was previously a file server, so 4GB was more than sufficient.
  3. Currently the Veeam server is 2003.  I’m thinking 2008 since we are starting from scratch.  Any things to look for there?
  4. What’s the recommended backup schedule?  Backup all at once?  Split up based on host/lun?  Limit backup to a specific number of VMs at once?

Thanks for your help.  I could figure out by trail and error, but would rather not recreate the wheel.  What works for you, and what is best practice?



5 Responses

  1. Hey Guys! Great outline here, my comments are:

    -For the best writes on the storage target; RAID 10 is best -> But that is a lot of overhead… (In terms of available storage) -> If you can do it, do it. If not, I’d recommend RAID 6.

    -Veeam’s I/O behavior during jobs has high I/O, both read and write. New jobs and incrementals are high write. Synthetic fulls are high read and high write. (another nod for RAID 10)

    -The RAM is “OK”

    -2003 is fine, especially if tape support is needed as NTBackup can do a native offload for your.

    -In terms of consecutive jobs, I’d zero in on your backup window. 23 VMs isn’t really that much. I would make two jobs: 1 for Windows VMs and one for Linux VMs. If there is a logical structure/separation of your VMs, make one job for each of those separations (Dev/Prod for example). Also consider making jobs run in parallel (and have the same name) as vSphere folders and folders on the backup target -> That self documents the flow.

    Happy backups, don’t hesitate to reach out for more questions.

    Rick Vanover, I work for Veeam Software /
    +1 616 987 0691

  2. Cisco- Just my .02… others will likely have a different opinion.
    RAID level: RAID10 is overkill… you’d be wasting HALF of your usable space just for performance, and you don’t need THAT kind of performance in a backup array. If it were my array, I’d either make it a 14 drive RAID6 with online spare, or 14 drive RAID50 with online spare. Both would give you 24TB usable storage… the trade-off being better write performance (RAID50) and the ability to withstand two simultaneous drive failures AS LONG AS those two drives happen to be in different RAID5 sub-arrays vs. lesser write performance (RAID6) and the ability to withstand ANY two simultaneous drive failures in the entire array. All that said, assuming online spare capability (your 15th drive), RAID50 would be my choice.

    Also, your backup window will likely be far shorter than you imagine… once you get past the initial backups, your deltas will happen pretty darn fast, especially if you combine similar VMs into the same job, thus taking advantage of Veeam’s de-dupe capabilities. I back up about 2TB of data every night to my QNAP (which is configured as slow RAID6) over a single GB link in about 90 minutes.

    Your server looks reasonable to me… my Veeam box is a 2850 dual 3.6 with 8GB ram and only half of that is used, but I also run Veeam Monitor on the box so that’s taking some resources, too.

    I’m running Veeam on 2008R2 without issue. Definitely recommend that over 2003! Regarding backup schedule, I have a couple specific machines (like Exchange and Arena) in their own separate jobs, but the majority of them are in a “misc servers” job, which backs up each machine in succession. There are big advantages (mainly de-dupe) to backing up similar-OS machines in the same job. I then schedule my four or five individual jobs so they don’t overlap… pretty simple.

    Finally, I also REPLICATE important VMs (again, Exchange and Arena) every 2 hours from the SAN to local ESX storage… this is my “oh crap, the SAN just died” backup… it’s nice that Veeam can back up and replicate the same machines at different schedules, and still maintain the correct changed-block-tracking for each. The every-to-hour replicas literally take 5 min or so.

    Hope that helps.

    • Good point on local replication! Some people do that to a DAS resource, and others do it to an Openfiler/FreeNAS on lowest tier disk. Good points, Jim!

  3. Awesome writeup and comments guys. I’m planning my virtual deployment for second quarter next year, so this is a huge help for a jumping-off point.

    Sounds like Veeam is a must-have for our deployment.

  4. Awesome guys! That’s exactly the information I’m looking for. I’ll keep you updated. I’m building the RAID now. I’ll have to move a few LUNs from this server to our new file server and recreate permissions and shares before I actually install and configure Veeam. Making progress!! Thanks!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: