The backup vs. disaster recovery line is blurring. Backup applications can now do more frequent backups and provide recovery-in-place features that enable more rapid recoveries. In addition, replication applications that tend to focus on disaster recovery or business continuity are adding historical storing of replicated data as a key capability. The storing of previous versions of data allows these products to provide point-in-time recovery.
Given these capabilities, should organizations try to settle the backup vs. disaster recovery debate by eliminating one of these applications?
Can modern backups win?
Historically, backup has been copying data, often once per day, to a secondary storage device. In an era of rapidly changing data sets and new threats such as ransomware, once per day is not frequent enough. Backup applications are developing mechanisms such as changed block incremental backups to reduce the amount of data they need to transfer.
Recovering backups used to entail simply copying all the data you need from the backup storage appliance, often involving tape, to the primary storage device. Tape gave way to disk-based backup devices, improving the time to find the data to restore, but still had the same time delay in transferring that data across the network. Modern backup applications allow the mounting of the backup device as a volume so you can access the data directly, which eliminates, at least initially, the network transfer.
Armed with block-level incremental (BLI) backups and recovery in-place restores, backup applications can deliver impressive recovery point objectives and recovery time objectives (RPOs and RTOs). But BLI backups can't run continuously; it can take 30 minutes to 45 minutes before the recovery-in-place volume is available. This means most backup applications still struggle to deliver anything less than four-hour RPOs and RTOs, which may be fine for the great majority of applications in the data center, but not all of them. In a backup vs. disaster recovery discussion, data backup may have a bigger role than ever. But backup will fall short for many data centers because most will have at least one application that needs less than a four-hour recovery.
Is disaster recovery the top technology?
Disasters are not limited to a natural event that destroys an organization's entire data center. A modern disaster can also include more insidious attacks like ransomware, which is when your data is still in the center physically, but a virus or some kind of bug in the data prevents you from accessing it. Disaster recovery also implies a different type of recovery focus. Instead of just getting the data back, it focuses on getting an application up and running so users can return to work.
Meeting these stricter demands requires a more continuous capture of data, often by some sort of replication product, either a standalone software application or a function built into the storage system. In extreme cases, the replication has to be synchronous, which means writes have to be acknowledged in both the DR and primary site.
Typically, replication products are only concerned with the latest copy of data and are not used for point-in-time recoveries beyond the most recent copy. But some replication utilities can also snapshot their replicas. Snapshotting provides a versioning capability to the replication product, which it can also use to provide a point-in-time recovery capability similar to what backup software does.
Like backup, there are limitations to what replication can provide in terms of point-in-time recoveries leveraging snapshots. Typically they can only store a few days of snapshots, while a backup application may store many years of actual versions of the data.
Most replication products do not provide built-in data efficiency, like deduplication and compression, so they will require at least a mirror image of the primary storage capacity requirements. Finally, most replication products do not provide a sophisticated search function like a backup application does. The result is that while replication can fulfill many backup requirements, it can't provide long-term storage and discovery.
While the capabilities of both backup and disaster recovery are encroaching on each other, there are fundamental differences. For now, at least, a backup vs. disaster recovery discussion ends in a backup and disaster recovery discussion. There is value in the expanding capabilities of each product type. They can work together to enable an organization to better meet a variety of recovery demands.
Server virtualization is converging backup and DR
Backup evolution includes continuous data protection
Advancements enable backup and archiving convergence