Skip to main content
  • EN

Hi, all.

 

I have question about how to perform large backups on a slow network.

 

My current infrastructure is : headquarters and the IDC where the physical servers(119 TB) are located are more than 100km apart, connected by a 1Gbps network. (This network is for business, and I don't have any plans to set up a backup network yet.)

 

Initially, I wanted to have the backup server and repository server at headquarters so that I don't mind if the initial full backup takes a long time, but if I need to recovery backups, it will take too long. (Single largest server is over 7TB).

 

The second way I thought of is to install the backup server and repository server in IDC and make a backup copy to HQ.
However, the backup copy is also downloaded at 1Gbps, so I think the business network will be disrupted on the day of copying the full backup data.

 

How would be the best way to configure it in my situation?
Will the 'WAN Accelerator' feature help?
Also, does this work on a local network?

 

I need your advice.
Thank you.

 

 

if you max out the connection, even theoretical to transfer over a 1Gbps network would take about 23 Hours for 10TB.


That is if it doesn’t fail, connections don’t drop, and you would be flooding that connection to the absolute max.  

 

Either increased network bandwidth or perhaps looking at the process and slightly redesigning it make more sense.   Keeping large servers close to their backup servers make more sense than users being close to files they may not access.  

 

There will come a point in time where you need to move a job, or do an active full. 

Bandwidth is obviously the constraint in this scenario. Specifically, your offsite copy. I’ve made worse work, but it can get somewhat painful if the change rate is high on the servers. In the past we have setup at the local site and put a repo @ the offsite for backup copies. In some cases we physically ship disk with the backup chains and remap for the initial seeds. If you run everything else over that 1G link you’ll also likely need to throttle Veeam or risk other services having network congestion. 

The Veeam accelerating mechanisms will not help much with a bandwidth of 1Gb and above.

 

If you have some space at tje production site, you can try if the copy job seeding helps you.

https://helpcenter.veeam.com/docs/backup/vsphere/backup_copy_mapping_auxiliary.html?ver=120

 

The initial backup will have most data, this amount of data could be seeded. All subsequent backups are much smaller, so you can copy them over the network.

JMeixner and eblack have some good points.   seeding or physically shipping disks is an option to get it started with the initial backup. It doesn’t however solve the slow link issue. 

 

Just be aware of the following.

  • Some days there may be a high change rate.
  • CBT Bugs do happen
  • Future features or updates may require an active full or new chains to be created.

Veeam accelerators may help, but may not. If you are a 9-5 shop, you could at least configure the jobs to run after hours and set up policies to throttle traffic during the day. 

 

Comment


Terms & Conditions