How to Maximize WAN Accelerator Efficiency

[itlv2015]

I thought this would be a great place to share my experience with Veeam and discuss the best way I should implement Veeam's Wan Accelerators into our environment. Currently all data is on the production ESX host. Veeam backup and replication is running on a VM on the production host including the backup proxy service. Veeam does a full backup the first Sunday of the month followed by an incremental every day after. I have a backup copy job that runs over our 100mbit WAN to our off-site data center. My backup set is about to grow from 2TB to 7TB, so I want to get the Wan Accelerators setup to maximize WAN efficiency. Below is a drawing of how my environment is currently setup:



Questions I have to maximize the backups:

1. Where is the best place to install the Wan Accelerator? Should it go on the same VM that Veeam backup and replication is installed, a different dedicated VM on production host, or on the backup host?
2. Is it to much to replicate my main host to my second host (both at the same time) PLUS have the daily backups stored on the NAS and then replicated over the WAN?
3. Does it make sense to create an additional backup proxy anywhere on the host and target sites?

Thanks again for taking a few minutes to look over my drawing and questions. I moved our organization to Veeam back in December after having yet another Acronis issue.

[Gostev]

1. WAN accelerator placement will not make much difference in your scenario.

2. Yes, this will work fine. To reduce load on the production storage, you may want to backup VMs to "Main site NAS" first, then replicate from these backups (new v8 feature) to "Backup ESXi host", and also copy these backups to "Datacenter NAS" with a Backup Copy job.

3. Yes, most certainly you need will need have a target proxy in the DR site for the replication jobs to use, and this server will also be a gateway server for your DR NAS.

Thanks!

[itlv2015]

Thank you Gostev for taking a few minutes out of your day to give that explanation. Your response actually raised a few questions on this side which I was hoping you or someone could answer.

1. Could you explain a little why the Wan Accelerator placement isn't a big deal for my situation?

2. Wow what a great thought, we never really thought your method of taking stress off the production storage. So I can understand fully, I currently do a daily backup job from production server to "Main Site Nas" just like you said. After that I am currently running a "backup copy" to the datacenter NAS. How can I tell if the backup copy job is pulling data from the production server or if it's just copying the data from Main site NAS and syncing that over the WAN?

3. Currently we don't have a Proxy at the DR site. I will set that up including the wan accelerator by weeks end.

If I take your suggestion and do a "backup copy" job from On-site NAS to backup ESX server, how do I actually turn on the VM's on the backup server if the main server fails? It's my understanding that you do your recovery through the Veeam server user interface, which if the host is down then my Veeam server is also down. I think this is where I keep getting stuck in this situation.

Thanks,

[foggy]

1. Could you explain a little why the Wan Accelerator placement isn't a big deal for my situation?

You should set up a pair of WAN accelerators, one per each location so that they could transfer data efficiently between them. Actual location of WAN accelerator within the site does not really matter, it just should be placed somewhere near the storage.

2. Wow what a great thought, we never really thought your method of taking stress off the production storage. So I can understand fully, I currently do a daily backup job from production server to "Main Site Nas" just like you said. After that I am currently running a "backup copy" to the datacenter NAS. How can I tell if the backup copy job is pulling data from the production server or if it's just copying the data from Main site NAS and syncing that over the WAN?

Backup copy job always uses backup files as a source, it does not touch production VMs.

If I take your suggestion and do a "backup copy" job from On-site NAS to backup ESX server, how do I actually turn on the VM's on the backup server if the main server fails? It's my understanding that you do your recovery through the Veeam server user interface, which if the host is down then my Veeam server is also down. I think this is where I keep getting stuck in this situation.

A common practice is to make a second Veeam B&R installation located at the DR host/location responsible for replication jobs. This will allow for automatic failover in case main instance is down. Alternatively, you can perform manual failover (just start the VMs using vSphere Client).