V8 best practices for remote backups

[Tomsyr]

Hi,
We have successfully deployed Veeam to multiple sites. We are now going to target all of our backups to a remote host for both off-site availability and long term storage. Is there good documentation available regarding best practices? All but one site is on V7, but we plan to migrate all to V8.
Topics we want to address are:
Initial seeding of data from each site (we can send a USB drive from each site)
Best way to take advantage of WAN acceleration
Reverse Incremental VS Forward Incremental
The way our partner in the remote datacenter will be performing backups to tape will be to do a snapshot of the data each month, then archiving for long term storage. It will be very expensive and labor intensive to do restores, so we will want to a very long restore chain in place to handle the users that want a simple file recovery.
We have the ability to run multiple instances of Veeam at the remote datacenter, and it will be part of our domain.
Thanks in advance!
Tom

[foggy]

Tom, as far as I understand, you already have local backups at each site, right? Then it is indeed a good practice to seed the remote backups using those local ones and then create backup copy jobs and map them to the backup files transferred from the corresponding site using USB drive. You can configure the required GFS retention settings for long-term archival.

Backup copy jobs are always incremental, so no choice between forward/reverse in this case.

Btw, what kind of link do you have to the remote location?

[Tomsyr]

Thanks for the reply Foggy - Yes, we have local backups as well, the remote site will be for DR and long term.
Our links are all VPN based, and small sites are at 3 mbs to larger sites at 50 mbs.

So the seeding process is to create a 'new' backup copy, send it to the remote DC via USB, then enable the WAN based transfer and let the incrementals add from there?
Is there a good method/plan for maximizing the global cache?

[Gostev]

Is there a good method/plan for maximizing the global cache?

Not sure I understand your question.

[foggy]

So the seeding process is to create a 'new' backup copy, send it to the remote DC via USB, then enable the WAN based transfer and let the incrementals add from there?

Correct.

Is there a good method/plan for maximizing the global cache?

Are you probably talking about recommendations for the global cache size?

[Tomsyr]

For the global cache, it's my understanding that it 'learns' what type of traffic is crossing over the wire, so I am wondering if I should use one site with the fatter pipe to NOT use a USB transfer, so traffic like windows OS files populate the global cache. If I only send via USB, the cache is not aware of the the data - right?

[Gostev]

Correct. But v8 supports manual cache population too, you can schedule that periodically instead.