HDS fixin' to bring out common storage platform
One VSP ring to rule them all
El Reg understands that Hitachi Data Systems wants to move to having a single storage platform that can cover both high-end and mid-range storage needs.
Discussions with HDS executives and Bob Plumridge, its chief technology officer in EMEA, covered the AMS mid-range storage area and what lies ahead. The point was made that having two separate storage lines was undesirable, as customers face a migration from one product to the other. It would be much better if, say, an AMS 2500 customer could simply swap out the AMS controller and replace it with a new one that enabled scalability into and up through the VSP high-end storage line.
The VSP storage array line uses Intel X86 processors whereas the AMS line does not. It begins to look as if the AMS roadmap includes a shift to X86 controller code and the execution of that code on the VSP. A new low-end or entry-level VSP would be produced which could be gradable to full VSP capability.
Were this to happen then AMS customers would gain, in theory, mainframe-attach, dynamic tiering, and third-party array virtualisation capabilities; three functions they do not have presently.
We understand that most AMS sales are made to enterprise customers, with some 80 per cent of HDS revenues coming from enterprises, customers who probably have VSP and the previous USP-V arrays as well. Having a single HDS storage platform would make life better for them in terms of storage management and support.
Making a VSP controller affordable for existing AMS customers could be achieved, in theory, by reducing its performance and/or functionality and making it upgradable. A side aspect of this is the development of common drive shelves for VSP and AMS environments.
This development strategy, if implemented in products, is not something that could happen in the next few weeks or months.
As a strategy it would enable HDS to leapfrog HP which had its own common storage platform strategy before the 3PAR purchase and which did not include the XP arrays OEM'd from Hitachi. EMC appears to have a strategy of moving to a single mid-range storage platform centred on the VNX product, although there are indications that the high-end VMAX array could run mid-range array storage environments as well as the VMAX Enginuity operating system, with them all running as virtual machines under an ESX hypervisor.
Neither Dell, IBM nor Oracle appear to have unified storage platforms in their development roadmaps. Neither does NetApp with its purchases of Bycast for object storage and Engenio for high-speed block array applications.
H blocks not
HDS has produced a reference architecture or template for use of HItachi 2000 Compute Blades, AMS storage and third-party networking switches in the Microsoft Hyper-V Cloud Fast Track programme, in which servers run Hyper-V and other Microsoft system software. (We won't call these H-blocks because of allusions to Ireland, prisons and lock-in.)
This is the first appearance of Hitachi servers and storage as an integrated stack in the HDS product range. Two more templates are coming. When HDS people were asked about similar configurations for Cisco servers and VMware they smiled.
There is no indication that Hitachi network products are about to make an appearance in these integrated stacks.
On a separate note the HDS people were not seeing any demand from their customers for all-flash arrays.
El Reg would not be surprised if HDS' unified storage platform strategy was announced towards the end of the year, possibly in early 2012. ®
HP Common Storage Platform?
I enjoy your work and oft times I rely on it for concise up-to-date information. You folks are tops at what you do.
Tell me more about this HP Common Storage Platform? I count Dot Hill OEM (P2000 MSA), P4000 (LeftHand), EVA (DEC via Compaq), and P9000 (Hitachi, Ltd aka XP) as four seperate and incompatible technologies for block storage prior to adding 3PAR for block storage.
If I want files services from HP I buy a ProLiant running Windows or Red Hat Linux to accomplish this. Other than the fact that P4000 and the file servers are made from ProLiants, where is the common technology here?
HP with its nine separate and incompatible platforms for secondary and primary storage is the least unified portfolio in the business.
Disclaimer: I am a NetApp employee and ex-Compaq/HP.
HP is the least unified storage vendor
As Johny DeepFreeze said, HP is, and has been the least unified storage vendor in the market.
DotHill, Lefthand, Dec/Compaq, HDS and now 3Par. No migration path, no unified management and no unified functionality. Now throw in windows nas heads, polyserve and 2 different tape vendors and their storage story is a real mess.
No storage vendor is unified
I'd say IBM were worse. So IBM have DS3000, 4000, 5000, 6000 (may have stopped now), 8000, v7000, SVC, XIV, nSeries, DS5000 (DDN), NAS heads and 2 tape vendors and some I have missed. Over the past few years they have gone through distinct phases - first SVC was the answer now what's the question. Then nSeries. Then XIV. And now V7000 is the 'trendy' one amongst the IBM salesforce.
Dell have Equalogic, Compellent, EMC, Ocarina, etc.
EMC have a right old collection of different arrays cobbled together.
Even NetApp who were probably the only people who could claim to be a unified storage vendor have decided to buy LSI and diversify their portfolio. Why, because they realise that it simply isn't possible to have one single storage architecture that meets all requirements!
Simple fact is there is no 'one size / style fits all' for storage. You buy into a particular part of the portfolio for a reason and then scale within that product. With the HP range, for example, it's not like you outgrow a P4000 and you have to bin it and move to a 3PAR array. Both ranges start small and grow massive. BMW don't just sell one car. They have a full range of cars to meet differing needs, and within those ranges they have a variety of engine sizes and spec's.