The topic as to what storage management features organizations really need on a storage array continues to be a hotly debated. In the last decade, we have seen a multitude of features propagate on storage arrays including disk striping, thin provisioning, and storage tiering just to name a few. But deciding which of these features are “nice-to-haves” and which ones are really “needed” in a virtual operating environment (VOE) becomes very difficult to make without a close examination of one’s environment.
There is little argument that something needs to be done in the area of improving storage utilization in today’s enterprise environments. Even now in 2011 many organizations still find that up to 70% of their storage capacity on their storage arrays is misallocated or underutilized.
By way of example, a storage architect recently told me that he agreed to a request from his neighbor to do “a little talk” with the storage team at the company at which his neighbor works. It was only after arriving at the company that he discovered the audience for his “little talk” encompassed everyone from the CIO down to the storage administrators.
But in talking with them he found out that this company was on a trajectory to spend $14 million over the next five years on storage. Further, once he understood what storage they intended to buy and how they intended to manage it, he estimated that he could help them save millions of dollars if they considered an alternative storage management strategy.
Needless to say, when he said he could “help them save millions of dollars,” that got the CIO’s attention and led to him being awarded a consulting gig. By the time he completed the job he ended up saving them nearly $7 million over that same five year period.
But what is important to emphasize here is that his plan for this company involved more than just the company buying storage from an alternative vendor. It called for the company to completely change how it managed its storage infrastructure as well as re-examine what features it needed on its storage arrays.
In this particular case, the company took a two-pronged approach to eliminate its storage misallocation and improve its storage utilization in order to drive down costs. Notably, this approach did not call for it to buy newer storage array with the latest and greatest features. Rather, it was just the opposite.
Instead of purchasing storage arrays that offer features like thin provisioning and storage tiering, it purchased Symantec’s Veritas Operations Manager, Storage Foundation and storage resource management software that was installed on all of its servers.
Veritas Storage Foundation was acquired to improve the company’s utilization efficiency while Symantec’s Veritas CommandCentral Storage was purchased to manage its misallocated storage. All of this company’s storage was then brought under the control of Veritas Operations Manager.
The net result of this decision was that the company discovered that up to 80% of its storage capacity was being wasted and could be returned to the free pool. This storage reclamation eliminated the need for it to buy additional tier 1 storage capacity as it now had sufficient storage capacity to meet its needs.
Further, because Storage Foundation offered features like thin provisioning, replication, migration and storage tiering, the company was able to use those features across all of its application servers and did not need to purchase storage arrays that offered these features.
Having now solved this company’s immediate problem of a perceived shortage of storage capacity, it freed up time for them to quantify and understand exactly what features the company needed on its storage arrays going forward.
Together they examined today’s newest storage array management features (thin provisioning, storage tiering, etc.) and then mapped out which of these features the company really needed to meet its application requirements as opposed to what features it thought it needed.
The end result was that they determined the features that the company really needed on its storage arrays were ones that have come to be historically associated with storage arrays such as their availability, performance and reliability. However it also found it needed built-in management features that integrate with the virtualization hypervisors to reduce management complexity in their emerging virtualized environment.
By establishing these as the primary features that it needed a storage array to offer, the company opted to use Nexsan’s SASBeast and SATABeast storage arrays. In this case, it found that Nexsan provided the features that it really needed on a storage array at a much lower price than competitive products.
Establishing exactly what features should be included on a storage array is likely going to be different for every company and every environment. But what should be the same at every company is taking the time to establish exactly what storage management features are needed and where those features are best implemented.
In so doing companies will likely find they can achieve what this company accomplished: they can drive up their storage utilization; drive down wasted capacity that results from misallocation; drive down storage costs; and, in the process, make their storage infrastructure easier to manage.