The importance of Virtualizing External as well as Internal storage
by Hu Yoshida on Mar 27, 2009
George Crump and I are having a conversation about doing more with less, driving out cost while increasing efficiency through the use of virtualization. We agree for the most part except where I say that virtualization should be able to enhance commodity storage. He does not think that is important to the enterprise. As proof point, this he holds up 3PAR, Xiotech, and Compellent as leveraging virtualization to deliver smarter, more automated systems.
These vendors provide solutions for dynamic and thin provisioning of storage capacity in a box. They are excellent examples of virtualization of a single vendor platform. In other words they do not extend or leverage their virtualization to deliver smarts to other vendor’s products. They do not leverage virtualization to provide dynaminc or thin provisioning or other services like replication, business continuity, or data migration to storage systems outside of their own boxes.
The USP V/VM can leverage virtualization to provide dynamic and thin provisioning not only to its own internal storage, but also to other storage systems that connect through FC Standard storage ports. It also can provide non disruptive, moves, copies, replication, business continuity, and migration of thin volumes across externally virtualized storage. Some customers have even virtualized a 3PAR system behind a USP V in order to leverage the USP V’s replication and migration capability.
I would say that the ability to leverage virtualization across external storage systems is important for driving out cost and increasing the efficiency of all your storage assets. In stead of looking for a limited ROI on an investment in dynamic provisioning of a single box, you can look at an ROA, a Return on Asset that enhances all your storage assets through an investment in a storage virtualization controller that can leverage dynamic provisioning as well as other storage services across your existing storage systems.
Since all these services can be provided by an enterprise storage virtualization controller, you can lower your costs by attaching commodity storage from any vendor that provides the best deal of the day.
We will continue to sell our lower cost midrange storage and will compete aggressively for the privilege of installing our storage behind the USP V/VM.
Wikibon users have seen clear and compelling use cases for virtualizing heterogeneous arrays, particularly in environments where there are lots of siloed midrange arrays coming off lease on a regular cycle, causing serious migration headaches. Many customers have established a virtualized tier 2 as a ‘default’ platform for all storage, requiring clear justification to place data on tier 1. This can take migration planning from many months down to days or weeks and improve ROA as you suggest.
I have to say that three or four years ago, I had expected the uptake of this approach to be even greater than it’s been in the marketplace. In addition to Hitachi there are a couple of other credible suppliers of this capability– certainly enough to justify the model.
I think the initial planning for putting in a heterogeneous virtualized storage infrastructure has caused customers to delay adoption somewhat– unless there’s a pressing migration requirement.