Virtual Storage Equals Real Confusion Page 2Continued From Page 1
Storage virtualization intelligence is virtually all over the place. Host-based virtualization software provides independence from the SAN interconnect and storage assets, but must be administered on a per-server basis. Storage-based virtualization can often leverage other vendor-specific features/functionality, but typically lacks interoperability for heterogeneous storage environments.
SAN-based virtualization in the form of appliances or switch-resident intelligence offers centralization of virtualizing functions, but needs redundancy to provide high availability in the event of box failures. The market is still sorting out the strengths and weaknesses of each approach and there is as yet no overwhelming customer preference for a single solution.
In its current usage, the term storage virtualization applies more to systems than to discrete physical components. RAID, for example, is an atomic form of storage virtualization in that it presents a complex of physical drives as a single logical entity. RAID masks the complexity of individual drives and their geometries on the backend of storage systems while enhancing performance and data recoverability.
Storage as a Matter of Policy
The capability to treat multiple storage arrays as a single storage pool, for example, enables automation of repetitive and soul-numbing tasks such as LUN management. Automation of storage resources, in turn, accommodates a higher layer of policy selection and enforcement. Policy-based storage virtualization can govern use of the storage infrastructure to ensure proper allocation of different classes of storage. Priority applications get priority class of storage, secondary applications get less expensive classes of storage.
Policy-based intelligence is also a prerequisite for the next layer of application-aware storage virtualization. By monitoring data types, for example, storage virtualization can fine-tune data placement to more accurately meet the requirements of specific applications. Video data gets written to the outer tracks of higher performance disks; online transaction processing receives best-in-class snapshot care.
Lastly, with so much brainpower somewhere in the SAN, it becomes feasible for applications themselves to dictate what their storage requirements are via APIs to the virtualization intelligence. The parfait of functionality that spans from the abstraction layer, to automation, policy-based management, application-aware and application-responsive virtualization pushes storage networking towards its declared goal of becoming a ubiquitous and easily managed technology.
Virtually Potential Calls for Real Patience
Of course, none of this is rich and comprehensive functionality is available today. The ultimate promise of storage virtualization is still hovering somewhere in its own virtual reality. Current storage virtualization products achieve sundry bits and pieces of the ideal solution, but the technology is in its infancy compared to its rich, mature potential.
The goal of a ubiquitous, vendor-neutral storage utility as described by Compaq's Enterprise Network Storage Architecture (ENSA) and other industry literature will take more years of evolution of both the storage networking infrastructure and virtualization technology to achieve. The current state of the art, however, does have productive applications, as evidenced by the continued adoption of storage virtualization products for specific services such as point-in-time data copy. The confusion surrounding storage virtualization reflects its dynamism as an emerging technology, and while customers may have difficulty selecting the right solution for their specific requirements, due diligence and aspirin are highly recommended.
Director, Solutions and Technologies, McDATA Corporation
Author: Designing Storage Area Networks, Second Edition (2003) (available at Amazon.com), IP SANs (2002) (also available at Amazon.com).