Veeam: Scaling out to Amazon S3


As a long-time user of Veeam, its almost a bit bizarre that I’ve not written anything about it yet… and the time spent at home has allowed me to re-look at our Veeam installations to see what additional value we can get from the more recent releases, and the SOBR feature is the one that stood out, so I thought I’d put together a little post on what SOBR is, why it is useful and what I did to implement it.

If you’re anything like every other company I’ve worked for, you will have numerous sites, numerous storage platforms and numerous backup technologies. One thing I’ve been refreshing my knowledge on recently is Veeam, specifically their Backup & Replication product, the one they are very well known for.

I’ve been using Veeam for many years, backing up to NFS and SMB shares, then to tape, then using AWS’ storage gateway for our VMware Cloud platform, but I’d not honestly spent much time with it since v9, released back in 2016. Last year I was lucky enough to attend VeeamON in Miami, and it was an enjoyable event seeing how the company was moving forward, and in particular around cloud storage for on-premise servers. The recent covid19 lockdown has given me more time to look into their scale-out backup repositories (or SOBRs) and how I could standardise some of our backups across the business. This is a little article on what I’ve done

What is a SOBR?

A SOBR or scale out backup repository (repo) is Veeam’s terminology for combination of repos that are categorised into 2 different types: performance and capacity. Performance repos are essentially local disk storage, which could be a deduping appliance, an NFS share or directly attached storage, whereas capacity repos are object storage such as AWS S3, Azure Blob or IBM Cloud Object storage.

Why bother with a SOBR?

If you want to ensure you have your backups offsite, off your network and held in a resilient place for long term retention, then why would you not. At the time of writing for S3 infrequent access storage costs $0.0131 per GB per month. So lets say you need 20TB a month, at the time of writing, that is less than £250 per month, less than £3k a year. You’ll struggle to get reliable offsite storage, easily accessible from your backup platform natively for cheaper.

Setting up a SOBR

I’m not going to go through the detail of setting up Veeam performance repos, because the likelihood is you already have a target for your Veeam backups, but adding S3 as a target is ridiculously simple, I pretty much followed this handy-handy help page:

Within minutes I was connected to my S3 bucket in AWS, and I was ready to set up my SOBR. In typical Veeam fashion, this was ridiculously easy. You give it a name, choose which repos are going to be classed as performance, then you choose a placment policy, this just defines where each backup type (full or incremental) are stored then you select your capacity tier. In my case, it was AWS S3, and I chose to move my backups to S3 after 3 days, but I could have copied all of them up there if I wanted to. I chose whether I wanted to limit the times data was copied to AWS and then I was done.

How long does it take?

I think end-to-end it took less than 2hours before I got data up into S3 – pretty crazy really! I’ve spent some additional time looking into changing how my Veeam jobs work and am now looking to see whether Veeam can replace some other legacy backups that still go to tape (yeah, old school huh?).

The wonderful Anthony Spiteri has done a few handy videos which demonstrate this far better than my ramblings do which I’ve linked below. I’ve also put a link to Andrew Zhelezko’s whistle stop tour of Veeam > S3 which is a great resource too. If you’re an Azure kinda guy, Michael Cade – another of Veeam’s technologist’s recently did a demo of this and is just starting to deliver some excellent content on his YouTube channel.

Anthony Spiteri Demo: LINK

Andrew Zhelezko Video: LINK

Michael Cade Video: LINK

Leave a Reply