Veeam impressions

I was a part of a project that implemented Veeam 6.5 on our production network.  We were desperately in need of a new backup as the old one was out of space out of date, and simply not good for disaster recoery in our environment

A quick over view of our environment:

- 10 ESXi Hosts managed by Vsphere 5. 4 hosts are esxi 4.1, 5 are esxi 5, and 1 is a stand alone host with esxi 5.

- 4 SAN's of various sizes.  2 Dell EqualLogics, 1 EMC VNX 5300 and an older HP I canot recall the model of.

- 110 virtual machines varying between Windows server 2003 & 2008r2 and many different flavors of Linux.

- 2 data-centers connected by 1gb fiber (soon to be 10gb).

I implemented Veeam 6.5 about a week ago now.  I installed the software on a single VM.  (Server 2008 r2, 2 quad core CPU's and 12gb RAM).  We also got a new HP P2000 SAN.  I created 3 Luns on it.  One 12TB, another 8TB and the last 3TB.  I presented the 12TB and 3 TB Luns to the Veeam VM.  The configuration of Veeam is very straight forward and their website has great videos that walk you through setting up the software.  I found Veeam to be much easier to setup than Symantec Backup Exec but not quite as easy as Acronis VM protect 7 or 8.  But the performance of Veeam blows either of those other two products out of the pond!

Veeam is all about its built in deduplication and Vsphere or Hyper-V integration.  We have roughly 10TB in use on all of our SAN's combined.  Veeam was able to take all this data and dedup it down to 4TB on the first set of backups of all the VM's!  That is very impressive!  In order to get the best deduplication results with Veeam, you have to group as many similar VM's in a job as possible.  I have done my best to group our SQL servers in one job, file servers in another etc...  Currently I have a total of 11 jobs running at various times.  I have a few jobs that I set to only run once a week on Sunday.  (Terminal servers, domain controllers, print servers etc...).

Another nice feature of Veeam is that it does what is called a "synthetic full backup".  This means that rather than taking a full backup every so often as you set it to, it will do forever incremental backups.  On a day that you specify, you can Veeam run its synthetic full backup.  This will take all the incremental backups that you have based on the period of your retention and compile them into a full backup.  This takes the work load off of your production environment of having to run a full backup!  In our case our file servers would run for 12 hours to do a full backup.  Instead the Veeam server itself does this job with the backups it has for the VM's in its repository.  Its really slick!

I will follow up after I have run this in production for a while but I already have plans to look at setting up a Second VM to act as a proxy for Veeam to put some of the work loads onto as I have had some performance issues already on the single VM I have for Veeam.  Also, I just read that Server 2012 has a dedup feature for its volumes you can setup!  This would mean after Veeam's deduplication, windows could take my iscsi volume for my Veeam repository and dedup it even further!  Veeam has been boasting impressive dedup results with this type of setup!

Cheers!