Snapshots are powerful tools for creating a primary storage data protection system, one that offers faster recovery and can even replace traditional backup altogether. But protecting against primary storage system failure, which would render the snapshots useless, requires a second storage system, and even a third for disaster recovery (DR). This is obviously a major expense when you consider the cost of buying another primary storage system.
One of the ways to make this more affordable is to use a software-defined storage (SDS) architecture that can leverage commodity components. But the requirements of this secondary system are more than just low cost, it must also be extremely flexible. The right SDS solution can provide this flexibility too.
A Stand-in for Primary Storage
The secondary system needs to become a stand-in for the primary, providing protection against system failure and to maintain performance for critical data sets when the primary undergoes a drive rebuild. This means the secondary storage system will need to have a performance profile that’s at least in the ballpark with the primary storage system.
And it’s conceivable that this role may evolve. For example, the secondary system could be initially implemented to support a single primary storage system, with performance adequate to carry the primary system’s load as described above. But later, its configuration may change to support multiple primary storage systems – a many-to-one setup that requires more capacity or more performance. Or it may assume other functions as well, such as supporting test and development or data mining, since it houses a current copy of the production data sets.
The third storage system, the one pegged to be the DR system, probably needs a different configuration altogether, since it’s unlikely that it will be supporting real-time compute jobs. This box needs to have the capacity to support a replicated copy of the primary data set(s), including all the associated snapshot data. The replication software needs to be bandwidth efficient, for obvious reasons, so the SDS system must often support WAN optimization too.
Buying three brand name primary storage systems to accommodate this data protection strategy isn’t the answer, but buying three different systems (a primary, a secondary and a DR system) isn’t either. The point is that the architecture chosen for this secondary storage system needs to work for the DR system as well, and be able to evolve as data grows or other uses are identified.
Software-defined storage offers the architecture to provide this flexibility, as well as the support for low-cost, commodity hardware that makes the whole design affordable. A software-only SDS solution like Nexenta can take that flexibility even further by enabling the use of different hardware platforms for the local secondary storage system and the DR system as well, eliminating vendor lock-in. In addition, these systems include efficiency features like inline data reduction and thin provisioning, plus multi-protocol support and single-pane-of-glass management for multiple systems.
NexentaStor also includes active/active clustering and asynchronous replication to support the DR system. But, as a pure software play, it can also be managed via API scripts to support the use of a third party snapshot and replication solution, for added flexibility.
Data protection that’s incorporated into the primary storage infrastructure using snapshots and replication is a powerful option for companies looking to improve on traditional backup architectures and provide more real-time data protection. But this requires one or two additional storage systems that are somewhat on-par with the primary. Now, with software defined storage technologies and commodity hardware, these additional systems are more affordable. But the right SDS solution can do more by providing flexibility and simplified management of these storage systems, and integrate the snapshot and replication software used in these infrastructures as well.