With most modern storage systems, dealing with a storage volume that has run out of capacity is relatively easy. The ability to expand the volume on the storage system as well as the operating system's ability to recognize that expansion is straightforward. The big challenge is what to do when the storage system itself is maxed out. This is an area where storage virtualization can play a key role.
Total saturation of a storage system is not limited to just physical disk capacity. It also can mean the saturation of systems performance capabilities or I/O bandwidth resources. Without virtualization, using traditional systems, you have two options available when you reach one of these three situations. First, you can buy another system and manage it separately. For most storage managers, managing two or three separate systems is now a fact of life and really not that big of a deal. However, as the number of storage systems grows it becomes increasingly difficult to manage multiple systems. Even in the two or three system use case, it is difficult to properly load balance I/O and capacity resources.
The other option with traditional storage is to buy a newer, more powerful storage controller. Fortunately, most systems today will allow you to replace the storage controller but continue to use the existing disk shelves. Most vendors call this a data in-place upgrade. Some storage systems have the ability for you to do this upgrade without any downtime. There is, of course, the expense of having to buy a newer, more expensive storage controller and essentially throwing out your old storage controller.
Storage virtualization minimizes the downsides of both of these options by allowing you to use multiple storage systems and manage them as if they were one. With storage virtualization in place you merely add another storage system, assign it to the storage virtualization appliance, and continue to manage as you did before.
Most storage virtualization solutions will allow you to transparently migrate data from one storage system to another. This allows you to move more performance sensitive workloads to the newer faster controller without experiencing application downtime.
Some storage virtualization solutions will even allow you to extend a volume across multiple storage systems. Most IT administrators are uncomfortable with the thought of having a volume stretch across multiple controllers, but these storage virtualization vendors not only claim that it works but that it provides better performance because you can leverage all the storage systems controllers for I/O operations.
What here is what is most interesting about storage virtualization as it relates to dealing with overwhelmed storage systems: It allows you to bring in what might seemingly be point solutions to address specific needs, but still have them managed through a single pane of glass within the storage virtualization product.
For example, you could decide to deploy a solid state disk appliance to address performance problems on specific application workloads. With the ability to move data in real time seamlessly to these appliances you could move workloads in and out of SSD as needed. This would allow you to maximize your SSD investment across the broadest range of applications when they needed the performance of SSD the most.
Storage virtualization solutions that provide a single point of management and a single point of data services across heterogeneous storage systems have had trouble catching on in the marketplace. We believe that one of the points of accelerated adoption of these solutions will be the performance delta between new high-speed SSD appliances and more traditional legacy hard drive-based storage systems. This allows SSD vendors to focus solely on performance, not have to create their own storage services (snapshots, replication, clones), and better fit into a single storage management interface.
Follow Storage Switzerland on Twitter