It was a blast attending Veeam Velocity a few weeks back. This was their annual sales kickoff and their team has a ton of energy.
At this event, they released their Veeam Cloud Tier capability and we’re excited at SwiftStack to have a tested solution available now for Veeam customers—allowing them to keep up with the hyper-growth of data by leveraging private and hybrid cloud storage. While Veeam Cloud Tier enables the use of infinitely-scalable and lower-cost cloud storage, in one session, the 3-2-1 rule was mentioned, which I had not thought about in a while.
For those who do not know what the long-standing 3-2-1 rule is, it states that you should always have 3 copies of your data, on 2 different types of media, with 1 of those copies offsite. It was brought up by a Veeam architect when discussing strategies for using a scale-out backup repository with cloud storage. He highlighted that sometimes just using cloud storage with Veeam Cloud Tier will not alone meet the 3-2-1 rule and that another copy may be necessary. This is because the storage serving as the performance tier will only hold data for a short period of time before it is tiered off to “the cloud” (see the gold circle in the diagram below). If there’s only one copy in the cloud, that could result in only two copies of the data—maybe even only one if you need to retain and protect data deleted or versioned source or versions for a long period of time.
Veeam stores data onsite for a short period of time (like 30 days) and then moves the data off to cost-effective cloud storage for long-term retention.
For the sake of this discussion, I do not want to debate if the 3-2-1 rule is well past its prime, but instead show how it can be easily achieved with a cloud architecture and does not need to be supplemented with another job/activity. If you have to make additional copies of your data, it somewhat defeats the purpose of moving towards “cloud”.
3 Copies of the Data
The diagram above shows that sometimes you may only have 2 copies of the data if the data is only stored for a short period of time on the performance tier. While this is a good example of a space-efficient cloud architecture, if the source system or public cloud bucket becomes unavailable, you could be left with only a single copy of the data.
With a hybrid cloud or multi-region private cloud design, the operational simplicity of cloud is retained but you now have an additional copy of the data to provide a high probability that two copies of the data will be available in the event of a disaster.
In this hybrid cloud example, only one data center is available, so the public cloud is still used for offsite protection—but SwiftStack storage is deployed on-premises to provide that extra needed copy. Preventing further operating complexity, SwiftStack manages the data across private and public resources, providing Veeam with a “hybrid cloud”. SwiftStack 1space stitches everything together into a single namespace, so Veeam only sees one storage repository.
In this private cloud example, two data centers are available, allowing one data center to protect the other from a major disaster. Multi-region SwiftStack storage is used where nodes of the cluster are placed in each data center. Veeam sees one storage system and SwiftStack automatically replicates data between data centers to keep data highly durable and available.
2 Different Types of Media
Back when I started my career in the backup and recovery space, two different types of media often meant disk plus tape. The historical reason for why you wanted two different types of media is that chances of data loss decrease when the data is stored in two different ways, preventing media-specific issue from causing data loss. Tape has a ton of operational overhead and recovery time (RTO) implications, so we’ve been trying to get away from it for backup for well over a decade. A common replacement has been cloud storage (like SwiftStack), which is significantly different than traditional disk.
Disk attached to a controller protected by RAID is significantly different that disk in a distributed storage system protected by an algorithm that distributes data across multiple failure domains. RAID vs cloud is different enough to meet this requirement.
1 Copy Offsite
Since it’s far from recommended to store all backup data onsite, all of our design recommendations at SwiftStack involve replicating the data to either a second data center or the public cloud. This satisfies the “1 copy offsite” rule, but is just 1 copy on cloud-native storage ok? When deciding if this is sufficient for your data, you need to understand how that copy is stored and protected on this type of storage. First off, cloud storage provides significantly higher levels of durability than traditional siloed storage because of the way data is distributed across the cluster of resources. For example, this simple design provides around 14 nines of durability.
Distributed cloud storage, whether private or public, is designed to sustain multiple drive failures, node failures, and even full site failures without the loss of data. Even if you only have a single copy offsite in the event the primary data center becomes unavailable, you can have a confidence that data is safe and protected as you work to recover the primary environment.