Software Defined Storage (SDS) abstracts data services from the storage hardware, eliminating vendor lock in. With SDS a data center can be a mixture of different vendors’ storage solutions and the storage systems can be the best of breed for each specific environment. Both of these factors should drive down storage costs. But SDS brings a lot more to the IT table than just reduced costs. In fact, it is the other benefits that should encourage IT professionals to do more than just learn about SDS, it should motivate them to begin implementation now.
Elimination of Storage Knowledge Silos
Almost every modern storage system comes with a core set of built-in features. Typically, this includes thin provisioning, snapshots, clones, caching/tiering and replication. The problem is that each storage system implements these features differently. This means that an administrator has to learn the nuances of each system or, more likely, each storage system needs a different administrator. It becomes expensive to have all these individuals trained on the solution. It’s also hard to develop cross-platform storage strategies, since no one person knows the capabilities of all systems.
SDS resolves these issues by creating a common feature set that spans every storage system in the environment. Thin provisioning, snapshots, clones, caching/tiering and replication all execute the same way and can be managed from a single interface. Not only does this mean that fewer people need to be trained on the solution, it also means that strategies can now span storage systems. For example, snapshots can be replicated from one storage system to another for long-term retention.
Best of Breed Without the Headaches
The storage I/O requirements of workloads vary. Some need very high peak performance, some need very consistent performance and others just need to store data as cost effectively as possible. While there are storage systems that vendors claim can handle this variety of workload types, those optimized for a particular workload are typically a better fit for that workload. They are the best of breed. But best of breed often comes with additional storage management headaches, forcing the data center to either compromise on capabilities to get a single system that can deliver an “OK” price/performance ratio or suffer through the additional management headaches caused by multiple best-of-bread solutions.
The horizontal data services that SDS delivers also enable the selection of best-of-breed storage hardware, but without the headaches of managing multiple systems. In addition, some SDS solutions will allow for the ease of migration of workloads between storage systems when the needs of the application changes.
SDS Leads to Better Availability
Keeping applications protected and available is an expensive and complicated process. Part of the reason for this is the cost of the storage infrastructure to support such an effort. For the highest availability two separate storage systems should be in place and data should be synchronously or asynchronously replicated between those systems. With vendor-specific storage systems this replication had to occur between two like, often identical, storage systems. But for most organizations that secondary system was seldom used, it was just waiting for a failure in the primary storage system.
SDS enables better availability by allowing the secondary systems to be less expensive, tier-2 storage systems, since SDS can replicate between dissimilar systems from different vendors. In the case of a failure of the primary storage system, servers can be mapped to the secondary system rather quickly and be back in production with little data or time loss.
SDS Does Lead to Better Pricing
Of course the potential cost savings of SDS can’t be overlooked. The ability to force all vendors to compete for the business every time a new project comes up or needs to be expanded is a powerful negotiating advantage for the organization that runs an SDS implementation. Enabling a best-of-breed approach without creating additional headaches also lowers costs since a system can be bought for the exact use case intended. Finally, a big savings is derived from reducing the management time associated with storage. A single interface to all storage systems with a common platform for data services should require fewer storage administrators and reduce errors.