Storage Virtualization:
Simplified Storage Virtualization
Imagine a world where storage management is simplified; everything works together; there are no configuration issues when upgrading, and vendors all cooperate to allow information to be easily stored and retrieved seamlessly. This was a reality in the mainframe world, but the deployment of open systems shattered these concepts. Today, most storage array vendors are more focused on selling their own arrays versus creating a solution that is easy and cost effective to manage.
One proposed answer to these incompatibility and storage management challenges has been virtualization. Virtualization, in simplistic terms, creates logical views of the physical storage; it makes the storage appear to be one large identical storage pool. With storage virtualization, a user no longer needs to know how storage devices are configured, where they are located, or what their capacity is. Users can then take advantage of competitive pricing and purchase their storage from multiple vendors transparently utilizing additional capacity as needed.
Today, storage virtualization can occur at multiple levels such as at the device level and volume level (see Table 1). The key element in each case is how the virtualization system maps the logical presentation to physical devices, and where this mapping is maintained and controlled. Other factors include performance, interoperability, and scalability issues. For example, network (SAN fabric based) virtualization is the most scalable and array (storage based) virtualization has interoperable constraints.
The first type of virtualization available in the market and the most common; it was not initially defined as a virtualization solution but a volume manager and/or file system. Host-based virtualization requires a software program to be installed on each host that intercepts requests for the file system or the logical Host Based volumes and maps it to one or more devices attached directly to the host (or maps it via a SAN). A typical example is Veritas' Volume Manager (VxVM) and File System (VxFS) running on the host, which can be controlled by a central SAN manager.
Virtualization Constraints
All virtualization solutions available today have some constraints such as interoperability, performance and capacity (see Table 2). This point is made by Fred Moore in 101 Storage Solutions. "Presently, virtualization at the SAN and NAS level doesn't effectively manage the capacity, performance, and inefficiency issues at the individual storage device level."
Also, virtualization solutions expect that classes of storage LUNs be already provisioned and available for mapping by the virtualization layer, much like a volume manager. If no LUN exists for that class, you need to exit the virtualization appliance and provision the LUN mapping it back to the appliance.
Beneath the virtualization layer, all devices are not equal. Properly managing performance of these devices requires that each type of device capacity be managed accordingly. This remains a yet-to-be-addressed area in the scope of traditional storage virtualization solutions, which is being implemented with a variety of schemes.
Storage administrators are adopting virtualization solutions to address the lack of interoperability between vendors. However, storage virtualization solutions today only shift the interoperability issues from the device vendors (arrays and switches) to the storage virtualization software or hardware vendor. This is exacerbated by the fact that APIs do not exist for any of the major storage virtualization appliances, which limits the management ability of independent third party software vendors. Storage software for functions such as backup, mirroring, disaster recovery, and distributed storage management may revert to specific vendor solutions, which today are as closed as the current suite of storage hardware vendor solutions.
Traditional virtualization tools available today have an inherent capacity limitation due to the fact that they take control of only SAN attached storage; the tools do no management of NAS and DAS devices. This single control point or single point of failure and the limitations on the types of devices that can be managed restrict the virtualization tools to management of a small subset of the entire networked storage environment.
Storage Operations Management - A Storage Virtualization Solution
A Storage Operations Management (SOM) solution recognizes the unique characteristics of individual storage arrays and pools but allows the user and/or the system to manage and provision them in a similar or identical fashion - in essence providing a virtualized view.