One of the big battles over the next five years in the data center is going to be managing the unprecedented growth of capacity. Features like thin provisioning, snapshots, and cloning are key capabilities in curtailing storage growth. Many vendors have most or all of these features in their systems, but while features may share the same name they don't always deliver the same capabilities.
As we discussed in our white paper "The Guide to Thin Provisioning" thin provisioning is the ability to dynamically allocate storage as needed. This saves you from having to buy storage capacity up front, and instead lets you assign it as the application consumes it. Key capabilities to look for with thin provisioning are how small can the dynamic allocation unit be and is thin reclamation supported? Thin reclamation is the ability to reclaim storage capacity from a volume after data has been deleted on it. Without reclamation, thin volumes become fat.
In some systems these features can also impact overall system performance. The process of having to allocate capacity on-the-fly, track active and inactive segments of data when snapshots are in effect, and the tracking of new data trees built off of old snapshots (as is the case with cloning) all can place significant overhead on the system. Look for storage system software and/or hardware that is properly outfitted to compensate for the extra workload.
Another feature or capability that we expect to be front and center in 2012 is primary storage deduplication. As we discussed in our recent video "Optimizing Primary Storage", deduplication should now be considered in the same mix as thin provisioning, snapshots, and clones when trying to curtail storage growth. Its performance impact, as is the case with other storage services, can be largely mitigated by efficient software design, proper feature integration, and hardware compensation.
If the performance impact can be mitigated, the return on investment for deduplication in the right environment--VMware being an ideal example--can be significant. We have seen as much as a 5X gain in capacity efficiency when primary storage deduplication is leveraged fully.
There are of course features that storage systems can provide that are not necessarily designed to deliver capacity efficiency. One that we spoke about in our last entry are the features that can optimize the use of solid state storage like caching or automated tiering. Other features may include new forms of data protection that either increase the level of redundancy or the speed of recovery. All of which we will cover as this series continues.
Follow Storage Switzerland on Twitter