Chris Mellor's blog entry about the 'The Storage Array Killing Fields' and Scott Lowe's entry on 'The Future of NetApp' touch on very similar areas; what is the future of storage array manufacturers and in Scott's case, what is the future of a specific array vendor but many of Scott's points can be applied to any number of storage vendors.
So what is the future of the storage array? Are storage arrays even relevent today? This is a question which is going to be increasingly asked as we move towards a commodity based approach; it has wide reaching implications for the storage array vendors. Of course, when we talk about the future; what horizon are we talking about? 2 years, 5 years or even 15 years?
I suspect the short-term future of the storage array, certainly the next 2 years is probably fairly rosey. Demand for storage is strong and the general herd are not yet transitioning to dynamic, commodity-based data centres. If you are storage array vendor, you should be able to continue to make hay. But I think that you only have to look at the high-end array market to see some interesting trends.
VMAX? VMAX almost stands by itself at the moment; there are no real competing products in the market today. Huh? What about HDS and IBM? Well, HDS are due a refresh of the USP and then I suspect that V-MAX has a real competitor in that space. IBM however, really don't want to sell you DS8K unless you are a mainframe shop. IBM are no longer that interested in the high-end array; if you are a non-mainframe shop, they will try to sell XIV and they will do some interesting things to do so.
I do not expect HDS, EMC or IBM to vacate this space until the mainframe dies but the reality is that for most IT shops, the high-end array is probably not where they want to be; many of them have not realised this and are still buying high-end arrays because they always have done and that is what they are comfortable doing. In reality; Clariion, AMS, FAS, XIV, 3Par and a whole raft of others are going to do the job for most purposes; probably 80-90% of the workloads which run on high-end arrays today, no longer belong there.
It is in this mid-range sector that we are seeing most main-stream development and some of the major new features such as automated tiering and thin provisioning have come from there. It is a very busy sector however and as Chris points out in his blog entry, if integrated stacks become pervasive there are going to be casualties and some of the casualties could be large.
But there is another emerging market; that is bulk commodity storage; 'Cloud Storage' if you will and I'm not expecting this to delivered as part of the integrated stack for some time. My reasoning for it not being delivered as part of the integrated stack is that it is not yet well enough defined as a concept. Customers are not really ready for it, delivering it as part of the stack adds very little value to the stack at present.
Arguably much of the data which sits on mid-range storage could probably shift down to bulk commodity storage but I think that there needs to be more clarity around the delivery model for bulk storage and how it will/should be integrated.
So there is a window of opportunity for vendors to innovate in this space but will the traditional vendors seize it? Or will they simply see 'Cloud Storage' as simply another gateway into their current cash-cow? Or perhaps paint the cow a different colour and call it something else?
I think we've seen evidence of all of these approaches from the traditional storage vendors and maybe their history prevents them from seeing that the delivery model of their product needs to change.
In fact, this emerging market is being ignored by a great number of these traditional vendors who are hypnotised by the current array market which as Chris points out is about to become a killing field. There are going to be casualties….move to somewhere where lots of people are not trying to shoot each other. The new field is getting crowded but there's still time to stake a claim.
