Showing results for 
Search instead for 
Did you mean: 

Migration of VMs Across Netbackup Appliances

Level 2


Looking to understand what options exist for a particular infrastructure design.  Each vCenter cluster is a self sustained rack/pod with network, storage and compute.  Currrently there is also a Netbackup Appliance in each rack and a master deployed that controls the backup environment.  We would like to be able to do cross cluster migrations (in between racks) but this essentially breaks the backup architecture and requires us to either manaully migrate the backups to the new destination rack (time consuming and manual) or start from scratch as it related to the back ups, which is not acceptable.


Am I missing something, is there a better way to do this either with the current design or potentially with a different architecture?





Partner    VIP    Accredited Certified


Please explain in more detail what you mean by "but this essentially breaks the backup architecture".

Is there a single master or two?



It breaks the backup architechture because it would require me to either copy all of the data from one appliance to the other (which is extremely costly and not worth it), or to essentially start over on the appliance in destination rack/pod not taking advantage of the historical backups that we already have.  It also means I will still have old data on the original applicance but a new full and differentials on the new.  The new will not use deduplification, and essentially doubles my storage use.  If I frequently move VMs between racks/pods, it becomes a huge waste of storage.



Hello RiaanBadenhorst,


Just following up...



Do you have a mix of Linux and Windows guests that are doing these migrations between racks/pods?  Because if that is the case, you should still see very good dedupe rates.

Is there any reason not to just leave the backups on a particular appliance?  Is the storage in each rack/pod not available to all the media servers?

A picture of what your current architecture looks like would help a lot.

Thanks elanmbx,

Perhaps dedup is the wrong way to frame it.  The problem is not the OS, but the data that is unique to the individual VMs and the differentials from the fulls.  Think SQL databases.  So we ultimately have to have two fulls backups in order to maintain the necessary retention for the workloads.  The storage is available to all media servers, but the connections between them is 1GB today (at least I believe this is the case, I will follow-up).  Backups themselves are through FC.  Either way to migrate the date between the appliances is too slow for any normal operation.



I would recommend that you get sufficient network bandwidth *between* the pods such that you don't need to try to move the location of the backup when you move the guest.  

Is this a possibility?  Are these pods physically located close to each other?