It seems that some of the big 6 vendors are in disagreement as to whether there is a real “all-flash” segment of the storage market, or whether all-flash is simply an extension of primary storage, as detailed in this article from Chris Mellor on The Register. EMC (via VP Chad Sakac) claims that all-flash is a distinct category in it’s own right, while HP (via Chris Johnson) claim all-flash can be achieved through extension of existing hardware, in this case their 3PAR StoreServ and high end XP devices.
If we look at both companies, we see a totally different approach in strategy when it comes to evolving their product offerings. EMC have evolved through acquisition including VNX (which was CLARiiON), VPLEX, Isilion, Data Domain and XtremIO. They have also followed similar paths for software too. Once acquired, EMC looks to maximise the value of the technology, pushing it to all of their customers and expanding the platforms through incremental improvements. Very little effort it placed on harmonising what are very distinct solutions, with the result that customers have to learn multiple sets of skills or buy even more software (like VIPR) to link the products together.
From a business standpoint, EMC’s approach is wildly successful; they use their strong and perhaps aggressive salesforce to push new products to customers, acquiring what they see as the best product in new and emerging markets. From a customer perspective, there’s less time spent by EMC providing offerings that simplify their IT solutions, which can in many cases create product overlap and confusion.
If we look at HP, the storage team have taken a different approach altogether. The 3PAR StoreServ platform has been expanded from its mid-range positioning to high end 10000 solutions and entry-level 7000 models. The architecture has been modified to accommodate flash storage or to act as a hybrid array. Customers can use 3PAR StoreServ across the majority of their requirements and supplement additional needs (such as mainframe) with XP and other lines such as StoreOnce. The StoreServ systems have also recently been updated to provide both object and file support too.
Now, one product can’t fit all requirements and 3PAR StoreServ isn’t a panacea for all storage ills. However approaching from a customer perspective, HP seems to provide a more customer-friendly stance than EMC.
Strategy Drives Marketing
So how does that lead us back to the decision on all-flash? In EMC’s case, acquisition strategy drives their marketing positioning. Rather than evolve existing platforms, EMC prefers to create new market segments against which they can sell new platforms. Having a new market to talk about drives a growth in business with little damage to their existing ones. HP on the other hand have chosen a marketing positioning based more on a customer bias (and one that’s arguably more business-focused), which means using existing products to the fullest extent.
Looking at the wider market, vendors have taken one of three approaches:
- All-Flash Arrays – build a new system from scratch that is focused around flash and designed specifically to use flash technology.
- Flash-optimised – these systems are built from existing platforms and amended to cater for the additional performance flash offers. The relative success of these solutions depends on how well the underlying architecture can exploit the additional performance.
- Flash-Enhanced – these systems use flash in new ways (perhaps as a cache or log structured file) to deliver higher performance and can also be delivered in all-flash versions.
Where existing systems have been amended, vendors have been able to keep advanced features such as local and remote replication, whereas new all-flash solutions have added these over time. Many start-ups have focused purely on using all-flash products and so we see discussion of an all-flash market, despite the fact that we also see new hybrid solutions being developed, that in many cases will meet customer requirements.
Marketing is not Requirements
We have to be careful to separate requirements from marketing and buy products that fit requirements, rather than find a reason to use a new product. IT should be driven by what the business needs, not the other way around. At the moment it suits the market to have all-flash as a definition as it helps drive customer awareness of new solutions. However the market is quickly maturing and raw performance isn’t always the main requirement; features are becoming increasingly important. In addition, as new technology, such as NVDIMMs comes into play, products will evolve to use this technology; those that can use it directly will see the most benefit.
From EMC’s perspective, we’re already seeing the start of a new category being created through the acquisition last year of DSSD. We can expect EMC to evolve their definition of the ecosystem and look to justify why yet another new product category is needed. Watch out for announcements around EMC World 2015. In the meantime, more information can be found about flash in the enterprise including category definitions in the January report from Langton Blue entitled “Flash In The Enterprise 2015” (chargeable content). More details are available on our white paper page – http://langtonblue.com/white-papers/ – vendors included: EMC Corporation, Hitachi Data Systems, HP (Hewlett-Packard Corporation), Dell, NetApp Inc, IBM Corporation, Violin Memory Inc, Pure Storage Inc, Solidfire Inc, Kaminario Inc, Nimbus Data Inc, Tegile Inc, Skyera Inc.