Backup Copy & Server Location

[zzzp]

Hi there,

Am using backup copy to backup jobs accross a VPN .. Works nicely, but having issues with speeds.

I read Backup Copy uses CBT, but will it use a proxy in the DR site?

So at the moment im trying to replicate say 4TB of backups accross a rather average internet connection. Theyve alI been seeded, and Veeam keeps up just (going all day which isnt an issue).. I have my BnR server in the main site, and the copied backup jobs in a DR site. What im seeing, is read speeds of around 8MB a second, and transfers at 3MB a second (have a 30Mb pipe accross to DR) .. These read speeds, what exactly is happening there, and are the read speeds limited to pipe between both the sites? If I had my BnR server in the DR site, which I read is a preferred method I assume to use fail over features (and maybe other reasons too?) , would I see higher read speeds at all, or am I just purely limited by pipe. I have tried CIFS and using the gateway in the DR, and tried using a Windows store, and seem to be getting around the same speeds for both.

Thanks

[veremin]

Proxy servers are not involved in case of backup copy job, since all data transfer happens between two datamovers running on top of source and target repositories. In case of target repository being a CIFS share, the communication will occur between source repository and gateway server chosen in the setting of CIFS repository.

Based on the description, the existing pipe does look like a major limiting factor ("bottleneck").

Thanks.

[zzzp]

Ok thanks,

So the alternative which would utilize things better would be to use Replication to the DR site? That way the proxy's would be used and CBT should work nicely?

[veremin]

Those kinds of jobs serve different purposes: long term archival (backup copy job) and best RTO (replication one).

Also, be aware that backup copy job is forever incremental in its nature, meaning that starting from the initial cycle it will only transfer changes, not the images as whole. So, not sure about your last sentence and comparison that you did.

Thanks.

[zzzp]

Thanks

[veremin]

You're welcome.

From my perspective, it stands to reason to clarify the underlying goal first and only then stick to one of the available options in accordance: be it long-term archival (backup copy) or shortest RTO (replication).

Thanks.

[zzzp]

Ok .. Essentially wasnt so concerned about RTO in the DR site, a days worth of recovery time is sufficient, but are currently lacking the diskspace to house complete replica's, so thought id use backup copies for now however they dont seem to keep up over the small pipe we have there, where as some important replica machines in the TB size over at DR keep up nicely with the proxy and cbt doing its thing.. You also mentioned backup copies dont merge, so theyre forever incremental? Which will be a problem so I think Replicas are the only solution. I think im understanding it right.

[foggy]

Are you using WAN acceleration in the backup copy jobs?

[zzzp]

No WAN acceleration unfortunately

[veremin]

Essentially wasnt so concerned about RTO in the DR site, a days worth of recovery time is sufficient, but are currently lacking the diskspace to house complete replica's, so thought id use backup copies for now however they dont seem to keep up over the small pipe we have there, where as some important replica machines in the TB size over at DR keep up nicely with the proxy and cbt doing its thing

How often did replication jobs run and how often backup copy ones did? Just trying to understand the difference.

You also mentioned backup copies dont merge, so theyre forever incremental?

The fact that backup copy job is forever incremental does not contradict with the fact that backup copy merges the oldest restore point with the subsequent increment, once retention is reached.

Thanks.

[zzzp]

Ok thanks,. Backup copy run every day, basically running all the time to be honest to keep up (backup to local storage go every night so theres always a restore point), and replica machines go every 4 hours accross to DR

[foggy]

Basically, provided the schedule is similar, the amount of data transferred between sites should be comparable both for backup copy in direct mode and replication. What are the bottleneck stats for the backup copy jobs?

[zzzp]

Ok thanks.. Network on the backup copy jobs is the bottleneck which is to be expected. Not to worry tho, all the data eventually gets there, just going 24/7

[zzzp]

Would I be right in saying there would be less data trasnfered using backup copy because the backups are compressed ? Compared to a Replication job?

[foggy]

Then using WAN acceleration is what could be recommended in this case to optimize the backup window.

Both backup copy and replication job transfer data in a compressed format, provided Veeam data movers are installed on both sides.