IBM Continues To Upset Conventional Wisdom On Storage

There is a natural tendency to want to create a sense of order out of disorder. Now large IT vendors often have overlapping product lines with different architectural heritages. That is especially the case in storage, where mid-range storage arrays and enterprise-class storage arrays have different lineages, but where each may best be used is not always clearly delineated. Even within a product family, products can be dissimilar. Analysts, in particular, may find this mystifying (aren't there efficiencies at all levels from manufacturing to sales from having a unified product line being lost?). Nonetheless, we soldier on trying to create some product order even when there isn't any.

However, the truth of the matter is that product diversity is probably a good thing as it prevents a vendor from becoming locked into just one way of thinking, offers customers choices, and may enable one product to benefit from market trends where another may suffer (i.e., don't put all your eggs in one basket theory). Despite these rosy thoughts, some seem to think IBM went beyond the pale when it acquired XIV. One reason is that XIV has a very different architecture than conventional storage arrays and, for some reason, many analysts have found it hard to understand XIVs. Another reason is that XIV does not seem to fit easily into the product continuum from small to large arrays. XIV tends to be considered appropriate for larger sized customers, but it cannot really be positioned to either the left or right of IBM's enterprise-class DS8000 storage systems. That two products occupy a similar space implies that one should replace the other.

What people need to understand is that XIV allows IBM to defy conventional wisdom in classifying storage. To understand that, first recognize that vendors who come from a server heritage, notably Dell, HP, IBM, and Oracle-Sun, typically do not sell storage off-platform. That is, HP tends to not sell its storage to accounts that have IBM server platforms and vice versa. Yes, there are exceptions, but even though a DS8000 could work perfectly with non-IBM servers, it simply does not happen as a rule. Of course, vendors that come from a storage heritage, namely EMC and HDS, intend to work with server architectures of every sort.

IBM is not focusing on accounts where a customer has already bought non-XIV IBM storage, which limits cannibalization only to existing IBM-storage accounts where the customer really feels that XIV has characteristics that cannot be met by standard arrays. Instead, it is targeting accounts with IBM servers, but non-IBM storage and accounts with neither IBM servers nor IBM storage.

In the first case, XIV gives IBM a fresh storage "face" to attempt to replace the storage incumbent. In these cases, traditional IBM storage arrays did not succeed, so maybe the XIV, will. In the second case, traditional IBM storage has little chance against conventional products from the incumbent systems vendor. Here, XIV's unconventional design may attract mindshare and at least have an opportunity to bid for a share of wallet.All that sounds very good, but how is it working out in practice? Well, the answer is that IBM appears to be laughing all the way to the bank as it is now battling in accounts where it seeks to displace another vendor as the storage incumbent instead of simply trying to fend off challenges to existing accounts. IBM claims more than 450 customers and 1000 units installed in the first year under IBM. IBM states that greater than 70% of the XIV units are connected to other vendors' servers, and that greater than 20% of XIV sales are to entirely new hardware customers to IBM. This business is accretive to IBM as its traditional storage products would not have likely been as successful in those accounts.

Overall, customers seem to understand what XIV brings to the table. For those of you unfamiliar with the architecture, here it is a nutshell: XIV uses off-the-shelf Intel X86-based commodity servers that act as unintelligent routers, commodity gigabit Ethernet switches that act as a backplane, and data modules made up of commodity servers with SATA disk drives that act as smart storage subsystems via caching, replication, and virtualization functionality.

XIV's distribution algorithm for data spread data across all of the drives in a system. There is no one-to-one mapping of a specific volume to a specific drive, as is often the case in other storage systems. All spindles are utilized, which increases performance and throughput, and there are no hot spots since one disk is not excessively used while others remain idle.
 
An administrator can seamlessly add or remove drives, modules, or racks in XIV systems. Virtualization is automatically built-in as the administrator deals only with logical, not physical, drives. Moreover, spare capacity is used instead of spare drives, as all drives are used, so when rebuilds occur they are much faster. XIV claims rebuild performance of 750GB in 20 minutes or 2+TB per hour. Scalability, reliability, performance, and ease of use are among the reasons that customers seem to find attractive in XIV.

With new XIV v10.2 release, IBM continues to put frosting on the XIV cake with a number of enhancements, including the following:

XIV fills what was a competitive hole in IBM's storage product lines, and is a product that could be used to compete for storage business where IBM was not the incumbent regardless of whether the account used the company's servers or not. Some analysts and others may have a hard time figuring out what strategic role of XIV fulfills within IBM, but it appears that many customers do not have that problem. Not surprisingly, customers care only about what a product does for them.

Those organizations seem to embrace the scalability, reliability, performance, manageability, and TCO of XIV. Strangely enough, many products from IBM competitors could justifiably make the same claims so there must be some eureka moment that customers experience when they kick XIV's "tires."  If so, IBM should bottle it for their other products. The latest v10.2 release adds some nice features, thickening the frosting on the XIV cake. But really what IBM is trying to do is to get those potential customers that have not tried XIV to do so under the theory of "try it, you'll like it."  And, if past success is any indication, then many who try it will buy it and XIV will continue to fill an important niche in IBM's storage product line.