Hewlett-Packard made a major announcement in December: their grid-based StoreServ 7000 based on 3PAR software. 3PAR has been selling enterprise-class storage for almost a decade and HP is building on that legacy with a radical take on enterprise storage.
HP calls it converged storage. The concept is simple: enterprise-class scale-out storage that is configurable to meet a wide range of application requirements. Flexible commodity hardware to mix-and-match to meet requirements, while configuring SSDs, hard drives or tape makes it easy to meet a wide range of price-performance ratios.
It is the enterprise analog of Google’s and Amazon’s cloud infrastructures whose scale, flexibility and cost are up-ending traditional IT. I’ve been looking for this from a major vendor since I wrote about the Google File System 7 years ago.
GFS breaks that model and shows us what can be done when the entire storage paradigm is rethought. Build the availability around the devices, not in them, treat the storage infrastructure as a single system, not a collection of parts, extend the file system paradigm to include much of what we now consider storage management, including virtualization, continuous data protection, load balancing and capacity management.
GFS is not the future. But it shows us what the future can be.
A choice, not an echo
The prevailing vendor model is the stovepipe: for every function within the enterprise there is a specialized product or set of products. These stovepipes are then glued together with a management layer that is supposed to manage everything from a single pane of glass – but never does.
Competition is stovepipe vs. stovepipe. The big iron arrays from EMC, Hewlett-Packard and Hitachi all compete, in theory, on availability, performance, support and cost.
The reality? Account control is often the deciding factor. Competition has stagnated over the last 10 years.
EMC is the most successful stovepipe peddler. Their technology publishing model – buy young companies that have crossed the chasm and sell the hell out of them – reduces their engineering risk at the cost of grievous incompatibility.
Storage as a layer, not a pipe
But as long as EMC can afford to buy the best – and they can – competitors are always at a disadvantage. But HP’s converged storage model changes that – just as Amazon has forever changed enterprise IT.
HP says they’ve already won 1200 customers for their converged storage line.
The StorageMojo take
No one is going to out-EMC EMC. They have the largest collection of best-of-breed point products in the industry.
Dave Donatelli, formerly of EMC, knows this well. Thus HP’s flanking move with converged storage.
Assuming HP can deliver a substantial fraction of Amazon’s benefits inside the data center, they will do very well. But the real magic will come when customers start interconnecting islands of converged storage to create a storage layer – like the network layer – in their infrastructure.
There are substantial customer political obstacles, but Amazon’s success has changed how CFOs look at IT. Smart CIOs are listening and responding.
Congratulations to HP and the 3PAR team for coming up with the only credible challenge to EMC in the last 15 years. I wish them every success.
Courteous comments welcome, of course. The friendly folks at HP flew me to Europe last year to get the announcement first hand. But I’d like a technical briefing from a 3PAR architect.
I think HP is making moves in the right direction when it comes to crafting a product for which their portfolio was deficient. I can’t help but still think that the model mimics that of EMC to an extent but without the big checkbook. What remains to be seen is how well will they execute with their convergence strategy. I know they are targeting the vast numbers of existing EVA’s in the field, but after that well runs dry they will have to compete on their merits. How many of those 1200 customers are really “new” customers or are they old EVA owners who have moved to “3Par” and thus new to that type of platform?
What still remains to be discovered is if their “converged storage” really is focusing on a new or different paradigm or is it just relabeling a stovepipe something else.
From what I have seen and heard the new 7000 series is a hit with existing HP customers, but there is plenty of competition in the space that this line targets, and there is frankly far more competition that is more nimble and agile in how they do business.
As Randy Bias says ‘cloud’ means doing things the way Google does things and this applies to storage as much as compute or networking. While the large vendors can imitate the technology, they can’t imitate the costs however, which the large web shops only achieved by using open-source software. Many enterprise CIOs may think their storage needs won’t be as large as Google’s anytime soon, but this feels like a belief that will be proven wrong sooner than they think and proprietary platforms, while technically solid, will be expensive hardware-based albatrosses around the neck of the balance sheet.
HP 3PAR is good and competitive with EMC but they lack a good NAS head / gateway and are therefore weaker than EMC on the convergence front.
I am watching what HP are doing with interest. I’ve had great experiences with plenty of their SME products going back to the MSA1000 (painfully slow, but never had a single problem). The Dot Hill manufactured products have also been excellent. However I always found it annoying that everything was different. EVA could not work with MSA and the interfaces were different.
They have lost their way in recent years, but by settling on 3par, they seem to have found a good path that they are happy to settle on. I agree with jie that they need to get a decent NAS gateway out there, even if they just put a skin on something like FreeNAS.
Converged Infrastructure was coined by HP back in early 2009 at their analyst conference before Donatelli arrived and was really an articulation of the underlying hardware architecture behind what is now often called software defined infrastructure. The principals are sound but the need for scale out based platforms to take advantage have changed with the advent of solid state. No longer do you really need a cluster to drive performance because you can use SSD as a cache for many application workloads.. For larger enterprises having multiple storage software stacks is still viable because they are large. However for smaller enterprises real convergence dictates that you also converge the software. Most storage companies are now building upon industry standard hardware. The key though is to ask what are they doing to enable customers to capitalize efficiently on solid state and what are they doing to cut down the software stacks you have to manage. When you blend industry standard hardware, SSD and HDD for performance capacity based applications and bring together SAN and NAS in a single storage OS, you are really driving convergence.
NAS lack was true till December 2012. Pls, check your info. I know several NEW HP storage customers coming from EMC installed base. HP 3PAR is really easy to manage. Check it on YouTube! Rgds.
Apparently HP 3PAR has better NAS head/gw than EMC. just google it: HP StoreEasy
Jie I’m not so sure you can level that at HP anymore. Industry standard Proliants with CPU, memory & PCIe expansion. WSS2012 64 bit OS with scale out clustering, SMB3, NFS4, block based variable dedupe, compression encryption etc.
Maybe the SAN/NAS combo isn’t a single SKU like VNX with Clariion & Celerra under the covers, but with HP’s approach you can scale block file or compute independently and still provision storage centrally. http://goo.gl/naMJq
OK, so they have just released a gateway (can’t see any NDMP support which is strange though I guess not essential) . I don’t find the configuration convincing to front end a 3Par (which is being touted as tier 1 storage) with something running Windows storage server ( I’m sure is OK in an SMB environment on its own). It looks very mix and match , if they weren’t going to develop in-house then using something like Nexenta would seem a better fit.
I did put gateway in my earlier post but regarding my main point this config will never converge will it?
Sure HP will still probably win a reasonable number of customers over time (they have to) but from what I’ve seen they have needed to be very aggressive on price to beat EMC.
Is the stovepipe the…
– storage vendor,
– it’s product lines,
– the technology underlying the product lines,
– the funnel through which customers pay for the storage products they buy,
– the cost of moving from one vendor/product line to another,
– thinking of storage as a point product to be purchased/managed as an abstracted layer?
NDMP is only required on NAS systems that don’t allow putting agents on them. Windows has the advantage you can run agents on them, you can run any backup and anti-virus applications native on them..and quite frankly outperform (WSS 2012 !) all the proprietary NAS systems out there. (take into consideration that on those proprietary platforms you need external servers for backup and anti-virus so a lot of performance goes to waste due to this). And a lot more flexible in terms of configuration (You can add memory as needed, add CPU’s , I/O cards, etc..try that on a NetApp or EMC VNX..)
Yes, 3par & WSS gateways is mix and match..but why not? With the new WSS 2012 you can even manage your 3par array from within the Windows server manager (yes, both file & block in a sngle interface), making this a true ‘universal’ storage platform…
Need more high availability? You want disaster tolerance? Multi-site environments? all that is possible using 3par & Windows Storage Server (you will need HP’s CLX software for the Windows geo-clustering)
In my books that is what an enterprise environment should be able to do..
“with something running Windows storage server” What you mean like EMC VNX runs internally ?
Jie The reason they needed to be aggressive against EMC was that until recently HP didn’t have a proper mid range 3PAR array and EMC’s sales tactic, knowing they would almost certainly lose the technical argument was to low ball the solution, whenever possible positioning a dual controller midrange solution (VNX). In the hope of disrupting the deal and convincing the customer they were on an equal footing, after all it’s always easier to up-sell than trying to down sell from VMax. In some cases the Customer actually went for this tactic, and it meant HP had to get very aggressive on discounts. Now HP have a true midrange solution with all the high end 3PAR features, their addressable market just expanded massively, and EMC may just find the tables have turned.