Remote Backup

[Robdigi]

I would like to create a remote Windows backup repository on the other side of a WAN link. Will multiple Veeam servers be able to use this repository simultaneously?

Also, I read on the forum that the preferred backup method for this scenario is forward incremental. Has anyone had any luck using reversed incremental remotely?

Thanks.

[foggy]

If you are talking about backup proxy servers, then yes, you can point several proxies to the same repository and limit the number of concurrent jobs targeted at it to control the load. But backup repository cannot be shared between different Veeam backup servers. I.e. each repository should be controlled by a single console.

Regarding backup method, I suppose you are confusing it with replicating latest backup offsite or offloading backups to tape (where the amount of copied data is valuable). It doesn't really matters which method is used for backup to remote repository, as in both cases you move only incremental changes, not the full backup file over a slow link. What is more important here is your target storage performance as reversed incremental method puts 3x I/O load on the target and is much slower than the forward incremental.

[Robdigi]

Thanks for the quick response.

I have several sites that each has a Veeam backup server. These servers are the backup proxy and repository at their respective sites.

I need each of these individual servers to send a remote backup to the same remote Windows server. Are you saying this remote server needs to be a backup proxy? I was under the impression that it should be setup as a repository.

[Robdigi]

After reading your post again I have a better understanding of what you were saying.

Would it be best for me to send the backups to a CIFS share on the remote server given my current configuration?

[foggy]

Well, you could actually create a repository on each of your individual Veeam backup server pointing to the same Windows server (possibly different shares). But note that in this case load control mechanisms will not work (as the number of concurrent tasks and data ingestion rate are counted within each console independently).

[Robdigi]

When you say "load control mechanisms" are you referring to bandwidth throttling?

[Gostev]

He refers to amount of concurrent tasks per proxy and repository, and also max data ingest rate per repository.

[Robdigi]

So, just to be clear, bandwidth throttling WILL work if I am sending backups to a CIFS share?

[Gostev]

No, throttling only works between data mover agents (proxy to repository in case of backup, proxy to proxy in case of replication). Writes to a share that does not have proxying server (a CIFS share based backup repository setting) cannot be throttled.

[Robdigi]

I tried to run a remote backup to a share, and the job failed after about 15 GB was transferred.

I read in the manual that the backup repository agent is also responsible for maintaining a stable connection over slow WAN links. I am guessing this is why my job failed.

We have a Veeam server at each site because we need each admin to be responsible for their own backups/restores. We only have one server which can be used as a remote repository. It was previously running Backup Exec CPS, and has a large storage capacity.

Is there anything else I can do to try to make these remote backups work properly?

[Gostev]

TCP/IP connection can be very bad as far as latency and packet loss, but it must not drop - this is the requirement for the current version. If the connection drops, the job will be aborted.

We do plan to add resume transfer feature down the road though, as we somewhat struggle with remote full backups across Atlantic in our own production. Incremental backups completely very fast with close to 100% success rate, however full backups take 2 days and thus requires a good prayer! So, full backups over WAN are indeed always a big deal, even though we very rarely have to do those anyway.

[Robdigi]

Should I try to send a full backup job to the remote site via snail mail, and try to run only the reversed incrementals?

It's my understanding that I would just need to rescan the share. Is this correct?

[foggy]

Yes, you can perform local backup, set up repository on your remote site, move backup files there, then rescan the repository to import backup files, and finally map backup job to these backup files to continue running incrementals.

[Robdigi]

Would it be possible to install a full copy of Veeam on my remote DR server, and use it to pull backups from each of the individual sites (each running a full copy of Veeam)?

Would the individual sites act as proxies for the remote site? The backup proxy would be on the remote DR site.

If this does work, would there be any licensing issues? Each individual Veeam server is currently receiving a license from Enterprise Manager. If I added the remote DR server to the same Enterprise Manager would it try to request an additional license for each ESX server it connects to (even though they are already licensed)?

I hope this makes sense.

Thanks for all the help.

[Robdigi]

Just to clarify, I am proposing the above solution so that I could use a repository instead of a share. That way I will have the throttling capabilities.

[foggy]

In a pull scenario you actually need no Veeam servers in each individual site. Moreover, as proxies cannot be shared between different backup servers, you'd better install Veeam B&R in your DR only and promote all individual sites' installations into backup proxies for it.

No issues with licensing in both cases, as only source hosts are licensed and the number of source hosts is not changed.

Though I would note, that if you use CIFS share as a repository for your backups in a push scenario, you still can leverage throttling between source proxy and target proxying server (that will then write directly to the share).

[Robdigi]

I need to use multiple Veeam installations because I need each site to have the ability to backup/restore their own servers. If there was an ability to configure roles this would not be an issue.

With the push scenario, could multiple Veeam servers use the same remote target proxy server/CIFS share?

[Robdigi]

It looks like you may have answered my question in your previous post.

So with my current setup, the only way to achieve throttling is to have a separate remote server for each site. Is that a correct assumption?

[foggy]

Actually, not. Throttling rules should work for each pair of "individual site's Veeam server - remote server" separately, as your individual Veeam installations are independent. I.e. all connections to the same remote server will be limited according to rules.

[Robdigi]

With the release of version 6.1 I have decided to revive this thread. I am still unable to find a solution for remote backups that utilize throttling in my environment.

Just to recap the situation. I have multiple independent Veeam backup servers. Each server acts as both the proxy and repository at each site. In other words, each site admin manages their own server. I need each server to send a backup to the same centralized Windows server over the WAN.

I read the following in the release notes for version 6.1:

“Backup proxy servers and repositories can now be shared between multiple backup servers.”

Just to clarify does this mean I can now utilize throttling in this scenario? Can I setup the centralized server as a proxy/repository, and have it used by all the independent Veeam servers?

As always I appreciate your help.

[Vitaliy S.]

Hi Robert,

Throttling rules and concurrent tasks limitation are not shared between the proxies/repositories, but you can now use the centralized server as a proxy/repository and configure your independent backup servers on the remote sites to use that server.

Thanks!

[Robdigi]

You say that they don’t share rules, but will the throttling work at the individual site level? If I set throttling rules at each site will it ignore the rules, and use the full amount of bandwidth?

[Vitaliy S.]

You say that they don’t share rules, but will the throttling work at the individual site level?

Yes, nothing has changed in throttling rules/max concurrent tasks logic since the last reply from foggy (see below), but starting from 6.1 sharing backup repositories/proxies between independent Veeam Backup server installations is now officially supported.

Actually, not. Throttling rules should work for each pair of "individual site's Veeam server - remote server" separately, as your individual Veeam installations are independent. I.e. all connections to the same remote server will be limited according to rules.

[Robdigi]

Thanks for the clarification. I thought that foggy was saying this would work with a separate remote server for each Veeam server. I now understand what he meant.

I am going to give this another shot. Hopefully using a Veeam proxy/repository instead of a CIFS share will help maintain a reliable connection over the WAN.

I appreciate the help.

[Robdigi]

I tried running a remote backup twice, and it failed both times. The first time it stopped after 2.5 hours, and the second time after about 7 hours. Both times I received the following error:

"Error: Client error: An existing connection was forcibly closed by the remote host Unable to retrieve next block transmission command. Number of already processed blocks:"

Has anyone seen this error before?

[Vitaliy S.]

This error is caused by the network connection drop, see other topics regarding this error:
http://forums.veeam.com/search.php?keyw ... bly+closed