WAN accelerator setup

[larry]

I have two sites connected with a 150 meg WAN that backup each other and are a DR site for each other. All VMware. Can a WAN accelerator be both a source and a destination? If so do I just make them the same size? If not can you have two WAN accelerators on the same sever?
With the new WAN accelerator it seems I should do all backup jobs local then do a backup copy to remote sites is this right? I would see no reasons to do remote backups anymore but all remote backup copies.

[Gostev]

Can a WAN accelerator be both a source and a destination?

Yes.

If so do I just make them the same size?

Yes.

If not can you have two WAN accelerators on the same sever?

No, and this is not required. Single WAN accelerator can be both source and target for any amount of jobs. You can deploy WAN accelerators in both many to one (Branches > HQ backup copy), and one to many ways (HQ > multiple DR sites backup copy).

With the new WAN accelerator it seems I should do all backup jobs local then do a backup copy to remote sites is this right? I would see no reasons to do remote backups anymore but all remote backup copies.

Correct, there is no point in remote jobs any longer.

[larry]

thanks for the quick reply.

Also thank the team who made the manual, yes we are using it

I have big hopes for the WAN accelerator, will let you know.

So far every thing is working 100 percent.

[larry]

At each site I have a physical Veeam server. Can they share backup proxies and WAN accelerators?

[veremin]

Yep, the backup proxy/wan accelerator machine can be utilized by different backup servers. Though, in this case backup servers will be totally unaware about each other. For instance, the settings assigned to a proxy by a first backup server (maximum number of concurrent tasks, for example) will not be populated to the other backup server.

Thanks.

[larry]

When running a BackupCopy job my bottleneck is source wan but at times target wan ? ( this is very first job ) The CPU on the source wan accelerator is 50 percent at times but less than 10 percent most the time. I would expect the target and wan to be the bolttle neck but not source. I gave the source 12 gig or ram which it is using very little. My plan is to test at night where I can give the WAN accel lots of resources. When going into production I plan on giving the WAN accel a whole esx server ( I have idle ones at night)

Should source ever be the bottleneck?

My main test will be the second run and nightly runs.


To access data in the remote DR site do I import the backupcopy file with the DR veeam server at the remote site? Will importing the file break the connection to the veeam server in the main office?

[tsightler]

With 150Mb it's certainly possible that the Source WAN or Target WAN could be the bottleneck. The WAN accelerator is effectively trading disk I/O for WAN bandwidth savings. The WAN accelerator in Veeam is highly optimized to reduce the data for links up to 100Mb. While it will work fine with faster links, you won't likely see a huge speed increase, but rather that the amount of data used is less. Of course, you can then scale this with multiple WAN accelerators, assuming you have the I/O to spare.

So if it's indicating Source WAN or Target WAN, I'd anticipate that you're seeing I/O for the WAN cache be the limiting factor. Since it sounds like these are VMs, take a look at the latency on the disk holding the global cache and let us know what you see. I don't know that throwing a lot of memory and CPU is really going to make it much faster, although memory might help.

Can you share any information about the number of VMs and the amount of data that you need to copy? It would be very interesting.

[veremin]

To access data in the remote DR site do I import the backupcopy file with the DR veeam server at the remote site? Will importing the file break the connection to the veeam server in the main office?

The files produced by backup copy job are reflected in the “Backups” node of your primary backup server, so, should need be, you can initiate any restoration activity from there.

Otherwise, if your primary server goes down, for instance, you will have first to import these files to a backup console of a server that resides in DR site.

From my perspective, importing backup data to a console of DR VB&R server shouldn’t break any connection or whatsoever, unless these files are accessed in the same time by two different servers. So, if you want to test this scenario, you would need to temporarily disable a given backup copy job prior to testing.

Thanks.

[larry]

thanks, more info and questions.
Site 1
Physical Veeam server, I virtual Wan and backup proxy
backs up 30 vms – 11tb with change 1tb daily
The data : 100k of new check images a day – need to be offsite.
2 SQL servers

Site 2
Physical Veeam server, I virtual Wan and backup proxy
Backs up 25 vms – 8tb change 1.2tb daily
The Data: Exchange and 2 SQL servers

I have backups running in both directions at the same time. (Converting all remotes to backupcopies from backups) Site 1 nightly run on Monday sometimes is still running on Tuesday when production starts, reason is on Monday we have 30 percent more checks from the weekend, without using backup copies.

On page 417 of the user manual the important message at bottom ”You cannot assign one Source WAN to several jobs simultaneously” , if I give the source accelerator more CPU can more than one job be sent?
If my Wan accelerators are both source and target at times then are they all only one job at a time? Will they just queue is sent more than one job?
When I link backup jobs to backup copy jobs that have many VM’s ,will the copies start on a vm as soon as the VM backup is done or does all the VMs need to be backed up before any copies start?

[Gostev]

The same WAN accelerators can be used in any amount of jobs.

WAN accelerator acting as a source will only process one task at a time, other tasks will queue.
WAN accelerator acting as a target can accept and process multiple jobs at one.

The copy will start as soon as the new restore point appears for at least one VM processed by the given Backup Copy job.