WAN Mode

[mplep]

Hi,

I've in-place upgraded a v4.1.2 installation to v5 and enabled the WAN mode in the two backups jobs I have. Should that feature kick in with the current jobs or only with new jobs? The reason I ask is that the incremental backup times look no different. I really hoped the reduced block size scanning would improve things. Perhaps a bottleneck exists elsewhere?

FYI I'm performing incremental backups in VA mode across a 100MB leased line from our vSphere 4 hosts with CBT enabled on about 28 VM's. The Veeam server has 4GB RAM, four dedicated cores and local storage.

Any suggestions welcomed, I'm a Veeam newbie.

Thanks

Mark

[Gostev]

Hello, you need to either create new jobs, or initiate full backup on the existing jobs (right-click the job, select Perform Full Backup) for the new setting to kick in. Essentially, this optimization affects how backup storage (VBK) is created. VBK creation only happens at full backup.

I guarantee that you will see huge difference once this setting kicks in. This will be especially obvious after looking at your daily incremental backups sizes (you will see 2-3 times reduction). This functionality was in beta testing in real customer environments since December last year, so we know very well what to expect here

Thanks!

[mplep]

Hi Gostev,

That's great news. On a similar note, will I see the same benefits with new replication jobs?

Thanks

Mark

[Gostev]

That is correct, there are exactly identical benefits with new replication jobs set to this mode as well.

[dadger]

How do we make replication jobs take note of the new setting (there is no 'perform full backup')...do they just pick it up and go?

[Gostev]

No, you need to create new jobs in case of replication.

[mplep]

Hi,

The documentation says "significant processing overhead" if WAN mode is used. Can you please elaborate on this and where this overhead occurs please?

Thanks

Mark

[mplep]

Hi,

I've been observing an incremental backup job (testing backups and replication). I can see that very little network bandwidth is used most of the time. The time taken is in processing during the CBT process. Does the SAN dictate how fast that happens?

Thanks

Mark

[Gostev]

Hello,

The overhead is resulted by the quantity of blocks that needs to be processed when smaller block size is used. This obviously affects both processing engine (a few times more blocks to hash for dedupe), and storage itself (a few times more I/O operations needed to write the same amount of data).

Correct, random read I/O performance of SAN dictates this mostly.

Thanks

[yankeestom]

[merged into existing discussion]

I think the person that set up my Veeam backup chose the incorrect target type. It is currenly set to Local, but the Veeam server is at a different location than the server it is backing up. Because the data is going over the WAN it looks like we should be using the WAN target option.

My question: If I modify the job and change it to WAN target will it need to start over and redo the full backup? Or will it still do the normal incremental backup tonight and merge it in with the rotation that is currently stored?

If it does a full backup I'll need to make the change over the weekend; if it can change on the fly and still do a regular incremental after switching modes I'll fix it for tonight.

Thanks!