So IBM has a new mid-range array in the form of the Storwize v7000; the question is did it need one and what happens to the rest of the range? In the midrange space, IBM now have two of their own arrays in the form of XIV and the v7000; they also have the DS5k which is OEMed from LSI.
IBM beginning to disengage from their storage OEM partners was eminently predictable; IBM have finally woken up to the fact that they need to own the stack and partnering was not going to allow them to steer their course in the coming 'Stack Wars'. However, the DS5k probably does have a part to place at least in the short term in some niches; the EXP5060 high-density enclosure certainly has a part to play in this.
XIV, though, is looking vulnerable and not for the reasons that I think some people expect me to state. XIV, simply does not make sense for IBM economically; it costs too much to make I suspect. But 'Bod, XIV is made from commodity components; it should be as cheap as chips to make.
Yes, XIV is made from commodity components but it's made from an awful of them to build an array; the v7000 is also made from basically commodity components in a custom form-factor and it uses a lot less of them to deliver better performance and more capacity. The Bill of Materials for an XIV is going to be significantly higher than for the v7000; this makes v7000 more profitable to IBM and/or will allow them to discount the v7000 more substantially.
The latter is going to be hard with the current XIV set-up which pretty much runs with their own sales-force; IBM will find themselves competing with themselves but I am sure that they will be happy to accept the higher margins that the v7000 could deliver.
Of course, there are some savvy customers out there who are going to do the same sums as me and are going to ask some hard questions; if you can supply XIV for 'x', why do I have pay 'y' for v7000 when it costs you less to build and supply the v7000.
Now, I know there are some readers who are going to mention that XIV is rumoured to be delivering some much awaited improvements such as improved inter-node connectivity and clustering; I've also heard that something along the lines of NetApp's PAM may also be delivered. You could stick an XIV full of SAS drives which would improve performance but increase cost; you could build an XIV based around 2.5" drives which might have been an interesting proposition, improving performance and capacity. But none of these is going make XIV any cheaper to make and any more an attractive option for IBM.
XIV's killer feature is it's ease of use and self-tuning capability but are users willing to pay a premium for this to enable the range to continue to be economic for IBM to make? Could IBM actually position the XIV as the AS400 of storage; the box you stick in the corner and forget about? I guess, that's possible but storage is a phenomenally price competitive market despite the attempts of people to pretend it isn't.
And the v7000 does not look exactly hard to use and configure; I'm also sure that IBM have learnt lessons about self-tuning systems from the XIV which can be applied more economically to it's more conventionally architected storage systems. I expect IBM to keep XIV around for at least another generation; it's an insurance policy at least but I'm not sure it's got a long term future in it's current form.
XIV could find a new lease of life though as a combined compute/storage appliance; that might make a lot more sense. Perhaps as a grid-based deduplication device? Or it might simply go down in history as this storage generation's Iceberg? An array full of interesting ideas which never quite fulfilled it's potential and in ten years from now; we'll be say remember the XIV, it had that feature first.
But ultimately, it won't be the users who will kill the XIV as a storage array; it's IBM themselves, it just doesn't make sense.