WAN Acceleration Information

[homerjnick]

I am looking to test this...we currently do replication to our DR site over a ten meg link and my replication jobs finish close to the end of my backup window.

Thus looking at the best way to optimize/accelerate replication.

I know I can get a thirty day trial of this (we already have Enterprise license but not Enterprise Plus) but what is involved with setting it up? I can't seem to find any info on it?

Are they just separate proxies at both sites? Or are all proxies WAN accelerators? When the trial finishes is there anything to be done apart from editing the job to not use them? Can my existing replication jobs just use the WAN accelerators if I set the job to do so (option is greyed out at present)? Will there be any impact on replication times if I have to revert back to Direct mode?

Thanks.

[CarlMcDade]

Hey Nick.

So the WAN acceleration is a role. You need to reploy the role at both source and target locations.

Essentially when deploying the role (Backup Console > Backup Infrastructure > WAN Accelerators) you define a cache. This role can be on your existing source and target proxies if you like.

We deploy a virtual disk to each proxy, and allocate for WAN acceleration. Add the role and then configure the jobs to utilise this WAN Accerlator Cache.

You may need to install the trial license to get access to the WAN acceleration tab first.

Additionally i have a real world use case for this. Customer did backup copy to offsite repository (not Cloud Connect) and was running at a 28 hour backup window. Implemented WAN acceleration and this drop to 45 minutes!!

Cheers

[homerjnick]

Thanks for the reply.

So I can reuse my proxies at both site to do this. How much storage on this virtual disk is required for the cache?

Is it easy to revert with no impact to the replication jobs and replicas?

[CarlMcDade]

No probs.

Yep the role doesnt need to sit on its own box. It can share the proxy role.

It depends on how many VM's you're replicating, and how much data. We started at 100GB each site and grew as we saw fit.

Yes, you can change the job. if it doesnt work for you, change the job to not use WAN acceleration.

Quick caveat. You need to build up the WAN acceleration cache, so you might not see any efficiencies until the cache has been populated.

[homerjnick]

Ah ok...that is helpful...thanks.

[CarlMcDade]

Good luck and let me know how you get on. My contact details are on www.carlmcdade.com if you want to take it offline

Cheers

[homerjnick]

Got my trial Enterprise Plus license.

Installed a WAN accelerator at both sites. I only have one backup repository...at source site. I have pre-populated the Global Cache of the source WAN Accelerator although I close down the window and cannot reopen it to see any progress.

Replication test jobs refuse to start because the WAN accelerator is not ready. Would have been nice to see that listed in "Running Jobs" so I know when it is finished.

Are there any implications to NOT pre-populating the DR Target WAN Accelerator? Since my backup repo is at my source site the target WAN Accelerator it is going to have to poll over the network to the backup repo to pre-populate the target WAN Accelerator's Global Cache thus I'm unsure if I should do it...especially since I do not know how long that will take.

[veremin]

Isn't populate session reflected in history tab (at the very bottom there is a tab called history; might be worth taking a look)? Thanks.

[homerjnick]

Good shout, it is in there and took an hour...so will run tonight for target WAN.

[veremin]

Good to know that my input was helpful. Should other questions arise, let us know.

[homerjnick]

Well I was unaware that the WAN Accelerator can only do one Replication job at a time. My replication schedule has overlaps in jobs where I have one job per VM and with only one WAN Accelerator at each site it will bottleneck replication.

I'm guess I'm going to have to match WAN Accelerator numbers with Veeam Proxy numbers.

[veremin]

Target WAN Accelerator accepts multiple connections, so you will only need to increase number of source WAN Accelerators.

[zoltank]

One thing to be aware of, the WAN Accelerator is designed to minimize bandwidth and not to maximize speed.

For example, I was replicating over a 100mb line, and my jobs finished MUUUUCH faster after I disabled WAN Accelerator.

[veremin]

On the links with high bandwidth (100 Mbps and above), direct mode might indeed work faster. If you still want to stick with WAN Accelerators in such scenario, consider deploying multiple source WAN Accelerators to saturate the link.

Kindly, check Best Practices document for more information.

Thanks.

[blithespirit]

WAN Acceleration is really designed for links less than 50Mb, above this you will find direct is faster as the overhead of the calculation is outweighed by the speed of the link.

Best practice is up to 4 to 1 connectors so 4 source to 1 target accelerator but you must make sure you have enough resource to accommodate the extra.

As mentioned previously have a look at the Veeam BP guide on Wan Acceleration for some pointers and base calculations: https://bp.veeam.expert/resource_planni ... ation.html