back to article HP at storage crossroads

In the rosy glow of better quarterly results from HP, one forlorn story of mediocre performance repeated itself. StorageWorks declined 3 per cent and EVA revenues fell 5 per cent year on year. Shortly after Mark Hurd became HP's CEO in 2005, the StorageWorks division was in a mess, with a history of what we called embarassing …

COMMENTS

This topic is closed for new posts.
  1. Anonymous Coward
    Anonymous Coward

    Good strategy

    A good strategy would seem to be a scale-out approach to storage with an increasing commoditisation at the disk level and the functionality provided at a storage server level. HP can drive huge economies of scale as it buys more disk drives than everyone else put together and have >50% market share in blades. A commoditised (and very low cost) scalable shared disk pool, with storage engines (read blades maybe?) running Left Hand, Ibrix, maybe an EVA-type firmware, maybe a virtual tape engine, etc, etc, sounds a good solution. Engines could be combined to provide more performance, engines could have their purpose changed as requirements changed.

    People don't buy mid-range or enterprise storage because they want it. They buy it because they have a requirement for storage with mid-range or enterprise characteristics. If you can give low cost components the performance, resilience and functionality of enterprise class storage without having to invest in large, inflexible silo'd solutions then why wouldn't you? Beats the crap out of the 'just buy a bigger pair of controllers' approach.

    1. Justin Bennett
      Thumb Down

      If your an SME

      Once you have a datacentre with several thousand servers (physical & virtual) then SAN is easily the best way to go. No need to physically manage thousands of disks here there & everywhere, improved resilience to failure as the load is spread.

      Then the ability to handle vast iops comes into play to, less SANs for more servers, or a SAN that can cope with vast throughput for large ERP systems... In either environment, local disks are no use.

      1. Anonymous Coward
        Anonymous Coward

        You're missing the point ...

        My suggestion wasn't that every server has its' own local drives, nor is that HP's suggestion from what I can make out. My suggestion was that by having large numbers of cheap SAS spindles, with performance or storage nodes in front of them, then you can deliver performance, functionality, scalability and resilience using industry standard, commodity building blocks. A SAN is a 'storage area network' not an array as you seem to infer ... you can provide shared, networked storage without relying on limited legacy dual controller arrays! Vast numbers of spindles delivers the disk-based IOPS. The ability to have potentially large numbers of storage nodes (picture a full C7000 maybe) then you essentially have a load of commodity building blocks acting as a 16 controller storage array. That would offer controller performance as well. All most storage arrays are anyway is a couple of servers and some disk in a box - EMC still have a couple of NT-based controllers, IBM have a couple of pSeries, etc. And the ability to run what you want on those controllers gives you functionality and flexibility. And if your requirements change then you change what you run on the controllers. Think how slowly storage arrays evolve over time and then think how quickly Blades and Blade connectivity evolve. And how quickly those developments could be brought into a storage solution if it was built using them. Open your eyes to this - cheaper components doesn't mean an inferior solution.

  2. Wanda Lust

    Industry standard

    Industry standard, an oxymoron.

    The opportunity in low end storage is to wrap HP servers with storage branding, i.e. sell more servers.

  3. Sam Paton

    HP = Meh

    EVA is crap, XP you might as well buy from HDS or from Sun for much cheapness (how long that relationship lasts is another matter)

    The lefthand stuff, not really used it so can't comment too much.

    EMC, HDS, StorageTek and god forbid IBM (replace the 8000 now please) have a better portfolio of products if you want to form an end to end solution from a single vendor. If you are happy to manage multiple vendors then the world is your oyster and there's nothing to make many people go "ooo let's go to HP"

    1. Sean 12
      Thumb Up

      EVA Family

      I have to reply to Sam Paton post........

      As far as I can see the EVA is a pretty solid piece of kit.

      Most of the PAS systems in the NHS run on it, without fault or failure!

      I know I've put them in!

      In Fiber land (which we all know is quicker than iSCSI)

      The EVA is much better (stabler) than anything else from anyone else. I've seen (14 years exp)

      Lets remember

      EMC hods of downtime for firmware upgrades! Iffy code as well!

      IBM is rebadged LSI crap!.

      HDS is way too expensive...

      DELL is too awful to contemplate

      If HP want the SWD division to be profitable, it's quite simple really,

      either get real when pricing up the kit and /or add extra features and make people want to carry on buying it.

      Why not port SANi/Q to the EVA? Just a thought!

  4. Anonymous Coward
    Thumb Down

    Oh great.

    "We have our top guys working on it... we think... "

    That's not saying much really is it? This is HP we're talking about......

  5. Anonymous Coward
    Thumb Down

    So there goes another European base

    So it looks as though HP's Bristol site is now doomed. Having scrapped almost all of the research, sold off the bulk of the site, storage is the next victim....

  6. David Halko
    IT Angle

    HP needs a little of OpenSolaris in their storage portfolio

    HP need to add OpenSolaris with ZFS into their portfolio... virtually unlimited storage pool & file system size, dedup, compression, encryption, fast kernel based CIFS, fast kernel based NFS, fast kernel based iSCSI, fast hardware accelerated crypto, unified security at the file system level for same file systems being used between multiple operating systems, superior reliability through block level checksum, superior reliability through RAID write hole avoidance, virtually unlimited snapshots, write level fslash acceleration option, read level flash acceleration option, superior performance management instrumentation built in, lower consumer cost, no development overhead, no license costs, clustering on the way...

    Where's the IT angle?

    More-for-Less with higher profits is better for IT consumers as well as the Supplier!

    1. Freakyfeet
      FAIL

      @ David Halko's blatant Sun advert

      Great idea but when exactly would we get the time to actually get the S7000 working?

      You can spout techie features all you want but with 1 hour+ failover and hardly an Active Directory installs working then it'll just be marketing noise.

      Oh yeah, that's all you have right now isn't it?

    2. Matt Bryant Silver badge
      Boffin

      RE: HP needs a little of OpenSolaris in their storage portfolio

      ....like a hole in the head! So, has ZFS got round that can't mirror between ZFS instances problem? Amongst others. In the meantime, hp have a much superior option in LeftHand if you want scale-out, and it isn't held back by the continued uncertainty of what Larry is actually planning for OpenSlowaris.

      Where hp does have a problem is the longterm timing. Just as the large SMP market is shrinking, so is the enterprise SAN market. With storage virtualisation packages starting to offer the same (and some would say better options) as traditional, monolithic arrays, sooner or later all the storage vendors will have to make a choice as to when they dump the enterprise and go scale-out "commodity" pooled storage instead. Personally, I think hp has dithered over this point to the detriment of the EVA offering.

      The interesting thing is hp's enterprise storage seems to be being eaten from below. The EVA is an SMB upwards offering, it should be taking sales from older high-end arrays as the EVA grows in capability to match them. I'm wondering if the drop off has been in the lower end of EVA4400 sales. I think hp made a mistake letting the EVA4400 go customer-installable as the first thing you hear about when you compare notes with other users is moaning about how the EVA4400 "didn't work" when the reality was the install got botched. It probably doesn't help that the cheaper MSA range is also SAN-capable and scaleable enough to overlap the EVA4400. Makes me wonder if the EVA4400 is the albatross in question.

      1. Sean 12
        Grenade

        Follow on to M Bryant post

        Dont know if your based in the US or UK?

        I have to agree EVA4400 customer installable is a big mistake....

        It's quite easy to drive the EVA now compared to the HSG/HSJ days gone by.

        Customers who buy their storage normally need the value add of understanding where the unit will fit in (in the scheme of things)

        This consultancy they dont get buying off the shelf....Go to your local HP Storageworks partner!

        Hand grenade......

        Badly deployed storage can be a real nightmare

      2. David Halko
        Happy

        RE: RE: HP needs a little of OpenSolaris in their storage portfolio

        Matt Bryant posts, "has ZFS got round that can't mirror between ZFS instances problem?"

        Not sure what you are trying to insinuate, Matt. From the very first releases of ZFS, any zpool could have one or more mirrors added to it... and the pool could be extended (far beyond 8 or 16TB most current file systems.)

        Multiple "instances" of ZFS are not required to get past any types of volume limitation issues, common to other technologies.

        Matt Bryant posts, "if you want scale-out, and it isn't held back by the continued uncertainty of what Larry is actually planning..."

        Scale-Out using multiple servers can be done with OpenSource ZFS via OpenSource COMSTAR today, no uncertainty in the future here... and automated using Sun Cluster.

        http://netmgt.blogspot.com/2009/08/multi-node-cluster-shared-nothing.html

        If free & open source ZFS is not good enough, more traditional clustering solutions are also available using QFS, for scale-out.

        Native ZFS clustering is on the way - every release of ZFS sees more Lustre features added - the continued integration of ZFS & Lustre is a beautiful thing!

        http://wiki.lustre.org/index.php/ZFS_and_Lustre

        Other vendors are building their storage solutions around ZFS, there is nothing keeping HP from doing it. Open Source is about Community. Solaris is not the limiting factor, to joining the party - HP can choose to join as well as anyone else!

        http://www.onstor.com/bhive/t/5/short_content.jsp?content_id=3163

        http://www.nexenta.com/corp/products

        A level of uncertainty does emit from FUD deliverers - but not so much from the Open Source ZFS community.

  7. Anonymous Coward
    Grenade

    re: RE: HP needs a little of OpenSolaris in their storage portfolio

    They have to do something. HP's storage is hurting and is offering nothing to differentiate itself. IBM is IBM, so they can get away with ho-hum storage. Sun at least has some differentiation with flash/zfs integration and analytics. EMC and HDS own the high end (I prefer HDS, but EMC is the historical leader). NetApp, ONStor, Dell, LSI, etc are also out there and offering more than HP. HP has no excuse for laying still in this market. They're making money, while many others are losing. Put that money into improving your sad storage!

  8. DJUNIX

    Glad we only got one EVA

    We purchased a EVA 6000 a few years back and were looking to purchase some more storage (work in healthcare so always buying storage). HP came out and did a dog and pony show (like every other vendor) and some management was convinced to go down that road. I liked some of its offerings but wasn't sure. In the end we decided to go with HDS as our storage vendor (we just picked up a couple of IBM XIV's though). So if we did purchase the EVA's, I wonder what their road map would be for us??

    So IF there midrange is in disarray, maybe they should look at purchasing some smaller fish who could revive their storage products (maybe 3Par).

  9. James 85

    HP Dropped the Ball on Storage a Long Time Ago

    It started with the EVA 4100/6100 mess that were an interim and stopgap replacement for the ageing 4000/6000/8000. These were then replaced - literally, no in place upgrade - by the mediocre 4400/6400/8400.

    HP sell MSAs for a third the price and no a lot less functionality than a EVA. Lefthand are the same price as the EVA. They constantly push EVA/Lefthand where they are not appropriate rather than MSA.

    They have poor software for replication and shapshotting.

    They really need define a storage strategy and then execute it effectively. To save time, this will probably require a purchase.

    Rather than compounding the purchase of Lefthand with another small niche player, I would recommend purchasing a bigger company such as NetApp. This would given them a very broad reach and appeal.

    They could then enhance the MSA range with some better software for SMB and ditch the EVA mess.

  10. Anonymous Coward
    Linux

    HP admits defeat...not surprised.

    Basically, HP is saying that their storage platforms are outdated, and people do not like them. The MSA and EVA are outdated, and Lefthand...no thanks. I'll take an Equallogic array any day of the week. What does this article mean to you? Dell is open and has the #1 iSCSI platform, and an outstanding line of Fibre Channel arrays as well.

    1. Anonymous Coward
      Anonymous Coward

      Stuff and nonsense

      I'm pretty sure HP are not saying their arrays are outdated. Just as they are not saying internal disk is the way forward as this article seems to suggest. People write such nonsense.

      The MSA is far from outdated - for an entry array it packs a punch, is cheap as chips, will serve FC and iSCSI native from the same controller and has remote replication (over IP natively) and snapshots built in. And will scale to nearly 100TB so not exactly small anyway.

      EVA is not revolutionary any more, agreed. It lacks the scalability of some other platform, and lacks some of the newer, more gimmicky features such as automatic tiering, thin provisioning, etc but these will be there for new and already installed arrays in months. When it has them, then combined with the things it already does and does very well, then it will be a decent mid-range solution again. Continuous Access is lightyears ahead of the remote replication software for just about everything else in the midrange. And the EVA is as simple as some of the people who post on this forum!

      LeftHand kicks Equalogic every time! Not even a close fight! Scalability, performance, resilience, functionality. And the Virtual SAN Appliance is genius! How many reasons do you want?

      I think people need to be wary of HP. They have been relatively quiet for a while but scale out storage is the future and noone will be in a position to deliver that better than HP due to their huge economies of scale!

  11. Anonymous Coward
    Anonymous Coward

    re: Oh great

    "We have our top guys working on it..." Didn't all of the top HP guys leave to Seagate about 6 years ago, now with Xiotech?

  12. Adam 61

    Ball Dropping Penguins

    James, Look just because MSA is cheap doesn't mean its always appropriate for every scenario, and of course HP will push EVA and Lefthand more because its their tech ergo they make more money on it and for the customer can also discount it deeper. MSA is a Dot Hill OEM ergo they make less money. I think you'll find the price diff between MSA and EVA is now much smaller, and when the next Gen EVA's eventually arrive i hear they are going to use a common shelf for MSA and EVA so the FC drives in the EVA's being the most expensive element that'll narrow the gap further

    "EVA replication is rubbish" - no not at all - its actually pretty flexible in the deployment options you can do - far superior to the MSA which has only just released a form of Asynch replication

    Actually its alot better than that offered by IBM's DS4000 range and HDS AMS (a HDS guy I know says this too).

    Snapshot poor - compared to what? its quite good and easy to use, the one thing it lacks is some good application integration (apart from MS apps via VSS) like NetApp does for Oracle and SAP. But you've got 4 different snap types to choose from including space efficient snaps to full blown clones.

    That said it doesn't dress up the wider issue that the EVA is ageing rapidly and HP need to get their bums in gear to get a relevant and credible competitive offering out in the market, its supposed to becoming but when???

    As for the PenguinBoys comments about Dell............now who's fibre channel line is outstanding? - that would be EMC's not Dells and who knows how long that relationship will last for! EQL is a good solution and has a lot of creedance in the market but equally so does Lefthand both have their pro's and cons.

    1. Matt Bryant Silver badge
      Boffin

      RE: Ball Dropping Penguins

      I've seen some NDA info on the new EVAs "coming soon", but I'm not sure if they address the single biggest problem I see in the current EVA range, namely that any upgrade between models is a controller swap. I don't understand why hp can't just make the controllers with pluggable modules to add the extra ports etc for adding fibre loops? Then you could buy one controller pair as an EVA4x00, when you need just upgrade them in-place to EVA6x00 and then finally to EVA8x00. Sure, if you upgrade from an EVA4400 to a 6400 or 8400 you can leave the current shelves in, but you have to re-jig the cabling, and often re-site the controller pair if you ordered the 4400 with less than nine shelves.

      Otherwise, EVA is still quite decent, and easily managed. The replication is good provided the link is good, the cloning/snapshot abilities are actually well beyond what we regulalry use, and the multi-OS support makes it popular for us, especially how it integrates with VMware and SRM. I suppose it just needs a refresh and a bit of overall storage strategy.

This topic is closed for new posts.

Biting the hand that feeds IT © 1998–2021