Sign in / up

back to article HP detunes Violin, tunes up 3PAR

HP has decided to curtail its reselling agreement with Violin Memory in favour of concentrating on 3PAR solid-state storage. HP has been reselling Violin Memory's v-31xx and v-32xx flash memory array products since July of last year via a Hardware Product Purchase Agreement (HPPA). We understand it has no intention of …

COMMENTS

House rules Send corrections
This topic is closed for new posts.
  1. Nate Amsden

    HP Violin

    I was going to comment on another article written recently that blasted HP for not having flash storage and was going to point out that HP had OEM'd Violin, but after looking further at their solution it seemed very limited as to what it could do - from a x86 side they supported one and only one server - the DL980, and they had some support from Itanium for that array. All in all a fairly lacklustre endorsement, not sure why they didn't bother to certify more systems - maybe the Violin stuff is so $$ that it just makes sense to have a really fancy DL980 to go with it I don't know.

    http://h71028.www7.hp.com/enterprise/us/en/solutions/storage-vma-series-memory-arrays.html

    Hopefully 3PAR comes up with some good flash stuff, I've been hammering on them every chance I get to make it better. The day 3PAR submits a SPC-1 test with a hybrid flash/disk array I think is the day that they will have figured out something really great. I don't think they would submit such a system unless it was really good (perhaps hence no SPC-2 from 3PAR).

    2 0
    1. Anonymous Coward
      Reply Icon
      Anonymous Coward

      Re: HP Violin

      Bound to happen.... HP paid so much for 3PAR that every storage opportunity needs to = 3PAR in order for them to justify paying $3 billion plus. They are going to need to do something though. IBM has Texas Memory. Oracle has a whole bunch of x86 servers packed with Flash in Exa.

      1 3
      1. M. B.
        Reply Icon

        Re: HP Violin

        I'm seeing exactly that lately, lots of 3PAR pitches where the Lefthand kit would be more suitable and cost-effective. Even the P2000 in some cases - I've seen some pretty wild discounts on the F-class starter kits just trying to get people on to the bandwagon. Most medium businesses and up are being pitched 3PAR when HP is asked to come to the table.

        2 0
        1. Anonymous Coward
          Reply Icon
          Anonymous Coward

          Re: HP Violin

          It may seem like 3PAR is the answer but what's the question but that is because 3PAR is the best storage array on the market, bar none, but you're right, it didn't address the lower end of the market so well so you did see HP having to shoe horn a higher class array down to a relevant price point to try and hit that spot. Expect that to change very soon!

          And the Violin r'ship never amounted to much anyway! Violin haemorrhaging staff, and all the low hanging fruit has gone.

          3 1
        2. Anonymous Coward
          Reply Icon
          Anonymous Coward

          Re: HP Violin

          If you're a salesman, and you're targetted on revenue, and a customer has say £50k to spend, and is talking to all the usual suspects, do you try and squeeze the 'best' solution off your truck into that budget or do just try and sell something cheap and cheerful? You sell the best box you can squeeze into the budget - it gives you a better chance of winning, you have the same revenue towards your target, and hold on - the customer wins too as he gets potentially a higher class of solution than he was originally expecting. Salesmen will try and maximise their chances of success, and if you have a product like 3PAR in your kitbag then you will try and position it wherever possible.

          3 1
          1. Anonymous Coward
            Reply Icon
            Anonymous Coward

            Re: HP Violin

            " the customer wins too as he gets potentially a higher class of solution than he was originally expecting"

            could also be written as "the customer loses twice because he fell for your sales tricks and then has to pay through the nose for extended warranty and support for the life of the missold product".

            3PAR uses different hardware to everything else HP makes (for the minute) which means that you pay a premium for expansion. They also currently don't have any unlimited licensing options meaning that not only will you have to buy your 1TB extra space but also the snapshot licence, the replication licence etc etc etc...You also have to buy support for those items.

            This is a perfectly acceptable approach for customers who can afford it, but midsize customers are better off going with Equallogic, NetApp, Compellent, even EVA in the longer term. 3PAR doesn't have any features compelling enough to justify the price tag compared to rivals.

            2 1
            1. Anonymous Coward
              Reply Icon
              Anonymous Coward

              Re: HP Violin

              Customer loses twice? How do you figure that? They get a more functional / higher performance solution than they thought they could afford, and I don't know why they'd then get stung for support costs as support costs are always purchased up front with the array. I can't remember the last array I sold that didn't have at least 3, but more frequently 5 years, support included in the initial acquisition. So where are the hidden costs to sting you then?

              And any upgrades will be at a similar discount to the initial purchase, so not sure where you get the idea that upgrades will be a rip off either. You're either a very niaive customer, or a very poor salesman, to be throwing these ideas around.

              3PAR has at least one pretty nifty feature - the 50% guarantee meaning that you could buy a much smaller array than any of the other vendors you mention. And 3PAR isn't even more expensive than a comparable NetApp or Compellent array anyway.

              You need to wander out in the real world once in a while - only a short sighted salesman or a very inept customer would ever expect upgrade discounts to be much less than the initial purchase.

              2 3
              1. Anonymous Coward
                Reply Icon
                Anonymous Coward

                Re: HP Violin

                "the 50% guarantee meaning that you could buy a much smaller array than any of the other vendors you mention."

                What? The 50% thin provisioning guarantee. In order to use it, you need to be on a system which does not have thin-provisioning. 3PAR doesn't have an advantage over anyone here. HDS offered the same guarantee the next day. I am sure IBM and EMC would make the same offer if they had less respect for the intelligence of their customers. It is not a unique advantage of 3PAR. It is just comparing fixed provisioning to thin provisioning. Every provider uses thin-provisioning at this point. Pure marketing.

                0 2
                1. Man Mountain
                  Reply Icon

                  Re: HP Violin

                  So I guess you haven't read the bit in both IBM and EMC's best practices where is says only use thin provisioning for workloads that can tolerate some 'performance variability'?? Whereas over 90% of 3PAR customer thinly provision everything. And no need to set up silly thin provisioning pools or reserve space ... blimey, so many rules and provisos when using thin provisioning with everyone else!

                  And yes moving from fat to thin will save you space, obviously but how many other vendors will guarantee that you can buy a smaller array?? HDS, EMC, IBM etc all need to land the data on the new array and then thin provision it, so you still need to buy a 100TB array if you are moving from a 100TB array. 3PAR thins on the way in so you only need to buy a 50TB array in the first place!!

                  Haters gonna hate, but 3PAR thin provisioning is light years ahead of everyone else's.

                  3 0
                  1. Anonymous Coward
                    Reply Icon
                    Anonymous Coward

                    Re: HP Violin

                    Man Mountain,

                    So basically, what you're saying is that the 3PAR product has all of the functionality of Dell's Compellent(100% thin provisioning, thin import) and the cost of 3PAR is higher than Compellent due to proprietary (non-HP standard) hardware components - and 3PAR is more restrictive because unlike Compellent, you have to choose the size of the box you think you'll need over the lifetime of the array - buy too small and you risk a forklift upgrade once you fill it up; buy too large and you end up with the higher costs on something that was oversized and underused. Yep, the old EMC model once again.

                    1 2
                    1. Anonymous Coward
                      Reply Icon
                      Anonymous Coward

                      Re: HP Violin

                      Compellent, ah yes I forgot they're really scaleable aren't they, well on paper at least but when you need 20+ back end ports to support all of the disk you advertise as scaling to, but only then have room for 4 front end ports, the Compellent looks more like an archive.

                      Tell me do you still have to resync the whole shebang if you have a blip in the network for sync mirroring ? How about having to disable those clever data progression policies when you add SSD. Or how the block size you choose severely limits addressable capacity on the array regardless of the number of supported disks. Thin import, yes you mean the bit that eats so many CPU cycles on Compellent that you need to do it offline.

                      I could go on about the huge advantages 3PAR has in availability, performance and scalability to name a few, but you get the picture. No wonder Dell initially passed them over for 3PAR.

                      2 0
                  2. Anonymous Coward
                    Reply Icon
                    WTF?

                    Re: HP Violin

                    Exsqueeze me, Meg Mountain: 3PAR thin provisioning definitely introduces performance variability*. Effectively you're converting sequential I/O into random I/O as you hunt around the disks for wherever your many 16KB blocks actually reside. It's a significant performance hit in real life.

                    3PAR thin provisioning _hype_ may be light years ahead of everyone else's but they're all serving the same lumpy porridge.

                    *Don't even get me started on 3PAR non-deterministic latency issues that I've experienced.

                    0 2
                    1. Anonymous Coward
                      Reply Icon
                      Anonymous Coward

                      Re: HP Violin

                      @googoobay Ooh do tell us about the non deterministic latency, please post you findings with full disclosure of your test setup and methodology. Without those, your comments are merely FUD.

                      Firstly 3PAR Thin Provisioning does not hunt around the disks for 16KB blocks, neither the underlying Common Provisioning Groups, Logical Disks or Thin Provisioned volumes work in that manner. If it did introduce performance variability as you state, then it wouldn't be part of 3PAR best practice, nor would over 90% of the 3PAR install base be using it.

                      What you appear to be describing is the sequential read after write penalty that has traditionally affected Netapp and other write anywhere file layouts. This isn't the way 3PAR works, not understanding such fundamental architectural differences doesn't really fill me with confidence that your comments or test results would have much validity.

                      2 0
                  3. Anonymous Coward
                    Reply Icon
                    Anonymous Coward

                    Re: HP Violin

                    Naturally if you thin provision in a traditional RAID system, you will have multiple hosts fighting over a fewer number of spindles. If you over do thin provisioning, you will not have the spindles to provide performance... which is why the "performance variability" in VMAX. How is 3PAR difference in this respect?

                    If we are comparing ease of use for thin provisioning, IBM XIV, for instance, is natively thin provisioned. You cannot not have thin provisioning with XIV, no performance hit because it doesn't use traditional RAID. XIV rules and provisos are inherently simpler than 3PAR, if that is what we are going for, because they don't exist.

                    "HDS, EMC, IBM etc all need to land the data on the new array and then thin provision it, so you still need to buy a 100TB array if you are moving from a 100TB array. "

                    Obviously not true in XIV either. If you want the slickest thin provisioning, it is XIV because there are no LUNs assigned to specific spindles groups. Everything is natively thin provisioning.

                    0 2

                    1. This post has been deleted by its author

                    2. Anonymous Coward
                      Reply Icon
                      Anonymous Coward

                      Re: HP Violin

                      """which is why the "performance variability" in VMAX. How is 3PAR difference in this respect?"""

                      You need to go read up on how 3PAR actually works before you comment. 3PAR is not a traditional RAID system, so your performance is not bound by how many spindles you have in a raid group. In fact there isn't really a concept of creating raid groups, the whole system is wide striped by default, this is part of the autonomic nature of 3PAR. Neither does it employ pools over the top of raid groups as in VMAX, pools just isolate capacity and constrain I/O, so why would they want that baggage.

                      """If we are comparing ease of use for thin provisioning, IBM XIV, for instance"""

                      I don't think we were comparing ease of use, but the only reason XIV thin provisioning is easy is because there's no option to turn it off. You wouldn't know whether XIV takes a performance hit for thin provisioning, because by your own admission there's no way to compare thin and non-thin as non-thin just isn't an option. In fact the XIV is so one dimensional that the only real option you have is to set the LUN size. I'm reliably informed unless your LUN size is a factor of 17GB then XIV will give you a different size anyway. 3PAR thin provisioning is selectable on a per LUN basis, but can also be the default for all LUN's, so if you can't be bothered selecting it on a per LUN basis, then you don't have to. I hear the XIV thin reclamation process is measured is in software and is typically measured in weeks.

                      """f you want the slickest thin provisioning, it is XIV because there are no LUNs assigned to specific spindles groups"""

                      Same for 3PAR I'm afraid, LUN's are not tied to spindles, (see you need to research more) you can have the same spindle serving multiple raid levels and LUN's concurrently and you can switch everything around on the fly. Yes 3PAR support more than a single raid level, but outside of the XIV and Netapp bubbles, that's seen as a good thing. They also support more than 180 spindles, multiple disk capacities and speeds unlike XIV's limited one size fits all approach. We also probably shouldn't mention the 45% max usable capacity of XIV and the dual drive failure exposure issue either, wouldn't want details to get in the way of good marketing.

                      Overall I'd say you've shown your usual ignorance of anything outside of XIV technology. You need get off that one trick pony of yours and have a look outside the XIV bubble.

                      2 1
                      1. Anonymous Coward
                        Reply Icon
                        Anonymous Coward

                        Re: HP Violin

                        3PAR implements what they call VVs, virtual volumes, which can contain multiple controllers and LUNs, over a set of physical drives. The physical drives use traditional RAID groups under the VVs. The only difference is that the VVs are mapped to the host instead of mapping the RAID groups directly to the host as the LUNs. 3PAR can run RAID 10, striped and mirrored... XIV uses a RAID 10 derivative, but it all has to be set up. Other than a slight different volume management software function, nothing new here. The RAID groups are just abstracted by the VVs, it doesn't mean they are not there. Still in the parity calculation and RAID planning world.

                        XIV:

                        "I hear the XIV thin reclamation process is measured is in software and is typically measured in weeks."

                        It depends on the utilization of the XIV. If you pull a host, the system 0s out the unused space and reallocates. This is a background process, so as to not waste cycles when the modules are busy reclaiming. If the XIV is constantly busy, unlikely, it could take days, but it is not an onerous process. This is purely to prevent any degradation though. It isn't as though the system slows down, like traditional storage, during reclamation... just the opposite, which is why it takes so long. The system is waiting for a period when the modules are free.

                        "We also probably shouldn't mention the 45% max usable capacity of XIV and the dual drive failure exposure issue either, wouldn't want details to get in the way of good marketing."

                        People make a big deal about the 45% usable space, but take a look at the average storage system, or even the well managed storage system, and see if they are doing any better. They are not, every storage system uses a large amount of capacity of RAID or another type of redundancy. Can someone spend all day getting 5% better utilization? Maybe. Is it worth their time? No

                        "dual drive failure exposure issue"

                        It doesn't exist. It is an EMC created myth. You can go yank two drives at random and nothing is going to happen. All data is mirrored across different modules and is contained in three drives. XIV will rebuild the drives in 30 minutes max. Can 3PAR, with its dual controllers... unless you add a whole frame, rebuild a 1 TB drive in 30 minutes? EMC made some hay for awhile with that FUD, but when some very large shops put 100 plus frames of XIV in and didn't lose any data ever, they needed to move along. You can say a lot about IBM, but that they make unreliable systems is certainly not one of them.... Ironically, if EMC loses two drives in one of their RAID 5 stripes, you do lose data... and it is much more likely that you will as a RAID 5 rebuild takes days.

                        "They also support more than 180 spindles"

                        Unless you have a volume which requires more than 256 TB of usable storage, which you don't, it should be fine. All of the arrays are managed centrally, they just don't share controllers across frames via proprietary ASICs like 3PAR.

                        0 2
                        1. Anonymous Coward
                          Reply Icon
                          Anonymous Coward

                          Re: HP Violin

                          ""XIV uses a RAID 10 derivative, but it (3PAR) all has to be set up""

                          Not true you create a Virtual Volume or LUN and the 3PAR system manages the creation and ongoing expansion of all of the underlying data structures. So yes Raid is there underlying the volumes for protection, in the same way as on XIV, but it's not something a Customer needs to worry about or manage, this is one of the autonomic functions of 3PAR storage. The system also picks the best level of availability for the volume based on the number of drive enclosures to ensure a volume can survive the loss of an entire drive enclosure and all disks. This is regardless of whether the underlying data protection is Raid 10 or Raid 5. Like I said, You need to do some more research before commenting.

                          ""If the XIV is constantly busy, unlikely, it could take days""

                          thanks for the rather wordy confirmation, a Customer of mine stated weeks to reclaim, but even days sounds bad considering 3PAR does this pretty much in real time, without affecting performance.

                          ""People make a big deal about the 45% usable space"" on XIV

                          I wonder why :-) well I do know Customers running 3PAR in excess of 70% usable, and if we take over subscription into account, then utilization rates jump dramatically into the 100's of percentage points, so you're not really selling me on XIV's efficiency. What about that minimum 17GB volume thing I mentioned, where all volumes need to be a factor of 17GB, is that true ?

                          ""It doesn't exist. It is an EMC created myth....All data is mirrored across different modules and is contained in three drives""

                          That's strange you mentioned you use Raid-X and that has two mirrors, not three as you've just stated above, also the 45% max utilization of XIV backs up the fact you only have two mirrors. I've heard from XIV Customers that yes this is a possibility but it's a statistically rare event and the 30 minute rebuild touted by IBM is designed specifically to reduce the likelihood of this kind of failure. I doubt Customers are making this up, so I can only assume IBM told them this at some point and tried to assuage their fears with the rapid rebuild argument. So given how XIV spreads the data, what does happen if you lose two mirrored blocks ?

                          ""XIV will rebuild the drives in 30 minutes max. Can 3PAR, with its dual controllers""

                          What's with the dual controllers on 3PAR, again you're showing your ignorance, 3PAR can start at two and scale to eight, but yes 3PAR also tracks used blocks, so it can rebuild only written data rather than the entire disk reducing traditional rebuild times and subsequent exposure times. I'm not discussing EMC here so I can't comment on their problems, but could you describe the performance affect the XIV 30 minute rebuild has on incoming I/O, I hear though the rebuild is short, a disk rebuild event is very noticeable to attached hosts .

                          ""Unless you have a volume which requires more than 256 TB of usable storage""

                          Why would anyone want a 256 TB volume, again you're showing how naive your understanding of storage really is, if you hadn't already shown how badly informed you were on 3PAR I would have assumed you were just trying to avoid answering the question :-). Solving capacity problems is easy, just throw some large disk at it. But performance is a different matter, to square that circle you need spindles and guess what XIV is limited to 180 maximum.

                          "Oh you want to go faster mister Customer ? the XIV is limited to 180 x 7,200 RPM drives, I'm sorry but you need to buy another XIV". :-( The lack of scalability is precisely why ""some very large shops put 100 plus frames of XIV in""

                          Give it up XIV is not even in the same class, but it is an architecture that just keeps giving :-)

                          1 0
              2. Anonymous Coward
                Reply Icon
                Anonymous Coward

                Re: HP Violin

                Yes, the customer loses twice - at least twice.

                Firstly, the features you talk about are nothing special. They may have been several years ago but those features are available in mid-range arrays like Equallogic, NetApp and Compellent as standard functionality at a lower price point.

                Secondly, no one cares about the 50% guarantee - it's a marketing guarantee which is largely meaningless in the real world. Even you stated that 3PAR isn't even more expensive than comparisons; ipso facto, it's not 50% of the cost.

                As to the customer losing twice. They pay the up-front costs on the array and then they pay through the nose for upgrades over the cost of upgrading on a lower-tier storage platform. Seriously, do most of your customers know what their storage environment will look like in 3 years? And you have the gall to sell them a 5 year warranty?

                For the vast majority of mid-tier customers, the frame-based stuff is old hat. These customer sare simply going to be better off scaling up and scaling out as their needs change instead of buying into an expensive box that they may - or may not - need to fill up.

                0 3
                1. Anonymous Coward
                  Reply Icon
                  Anonymous Coward

                  Re: HP Violin

                  Yes some of those features are available on other platforms, the difference is they're bolt-on not baked-in like 3PAR. Whenever you bolt on, you end up with mutually exclusive features, take a look at Netapp & VNX as examples, lots and lots of features but equally as many caveats in their use.

                  As for your argument about the Customer losing twice, quite frankly it's pathetic, Whenever a Customer buys a solution they make sure they understand ongoing upgrade costs, it's a standard part of the sales cycle. If a Sales guy does mis-sell in this way, then He'll never get repeat business, so what' would be the point.

                  As for support if you're selling a solution and not just a point product then you're sizing it to scale through the 5 years so why wouldn't a customer ant to take the 5 year support upfront. It's also the time when he's likely to get the deepest discount on support, so to have a known 5 year cost upfront is beneficial instead of trying to renegotiate costs annually.

                  2 0
                2. Anonymous Coward
                  Reply Icon
                  Anonymous Coward

                  Re: HP Violin

                  """For the vast majority of mid-tier customers, the frame-based stuff is old hat. These customer are simply going to be better off scaling up and scaling out as their needs change instead of buying into an expensive box that they may - or may not - need to fill up."""

                  I think you're confusing 3PAR with VMAX, VSP, DS8000. The difference being 3PAR is not a monolithic frame based array, it's modular even at the high end, offering both scale-up and scale-out options in the same box, so you can take your pick. If you're talking Equalogic or Netapp then it's well known that they breed like rabbits in the datecentre precisely because they don't scale.

                  So do you buy a new array every couple of years as you're suggesting and start with implementation services, planning and downtime to migrate workloads etc, burning additional ports in the fabric, dealing with multiple support contracts and licensing agreements etc whilst creating lots of storage islands and stranded capacity along the way.

                  Personally I'd prefer the ability to keep scaling in modular increments without stranding capacity and being able to manage the whole thing in a single namespace.

                  2 0
            2. Anonymous Coward
              Reply Icon
              Anonymous Coward

              Re: HP Violin

              Pretty much any Customer I have ever dealt with will want to understand three things as part of any deal. The products licensing model, the included support and any uplift costs and the minimum discount levels for ongoing upgrades.

              You're suggesting Customer's are too naive to discuss or request any of these details upfront from the sales team which is totally unrealistic. If they don't request these details then they shouldn't be in charge of signing off a sick note never mind a budget.

              As the above comments suggest a like for like Netapp, EMC or Compellent after discounts have been applied (no one buys at list except your naive Customer example above) will be pretty much on price parity with a 3PAR array as they all operate in the same market segment.

              3PAR has many very compelling features, you either don't understand that or they just haven't clicked with you yet, but that's ok most people are initially skeptical about new ideas and things they can't get their head around.

              "The horse is here to stay but the automobile is only a novelty, a fad"

              " Heavier-than-air flying machines are impossible"

              "There is no reason anyone would want a computer in their home."

              1 0
              1. Anonymous Coward
                Reply Icon
                Anonymous Coward

                Re: HP Violin

                "3PAR has many very compelling features, you either don't understand that or they just haven't clicked with you yet, but that's ok most people are initially skeptical about new ideas and things they can't get their head around."

                Like what? What is it that 3PAR can do that, for instance, VMAX or VSP cannot do? It is much better than what HP had, nothing, in the storage space, but hardly anything revolutionary or markedly different than what everyone else is doing at the high end. Federated controllers, thin-provisioning, wide stripes, etc... at least EMC, IBM and HDS.

                0 2
                1. Anonymous Coward
                  Reply Icon
                  Anonymous Coward

                  Re: HP Violin

                  " Like what? What is it that 3PAR can do that, for instance, VMAX or VSP cannot do? "

                  Of course they can all offer the "me too" features, but as you should know "the devil is always in the detail". Understand the architecture, start reading the best practices and talking to the support guys and you'll soon find out they're supported, but no recommended or just completely unworkable given the underlying architecture.

                  2 0
                  1. Anonymous Coward
                    Reply Icon
                    Anonymous Coward

                    Re: HP Violin

                    "Of course they can all offer the "me too" features, but as you should know "the devil is always in the detail". Understand the architecture, start reading the best practices and talking to the support guys and you'll soon find out they're supported, but no recommended or just completely unworkable given the underlying architecture."

                    In other words... not much.

                    0 2

                    1. This post has been deleted by its author

                    2. Anonymous Coward
                      Reply Icon
                      Anonymous Coward

                      Re: HP Violin

                      "not much"

                      We could go on for ever :-), but if you don't believe there's much between the old legacy monolithic arrays and 3PAR, then what possible excuse other than mainframe, could you have for not selecting 3PAR over the legacy 20+ year old architectures.

                      1 0
            3. Anonymous Coward
              Reply Icon
              Anonymous Coward

              Re: HP Violin

              "could also be written as 'the customer loses twice because he fell for your sales tricks and then has to pay through the nose for extended warranty and support for the life of the missold product'"

              Exactly, it is the old EMC Sym pitch.... Look what a great deal I got for you on this DMX/3PAR box! 3 months later: That software feature will cost $100,000, expansion will cost 3x the price it would have cost in the mid-range system, etc.

              Yes, 3PAR is better than EVA, XP, all of their other arrays by a mile... which is why the big push, but it is not required for the majority of their mid-market users (big shops primarily use EMC, IBM and, occasionally, HDS) and will end up being more than they wanted at some point in the future be it an expansion, license, or refresh that hits them.

              0 2
    2. Lusty
      Reply Icon

      Re: HP Violin

      I don't know for certain, but it could very well be down to bandwidth.

      The Violin box is marketed purely on IOPS and latency. If you hook one up to a DL380 then these are too limited to saturate a single Violin shelf and so customers may complain that they are not "getting the IOPS". I've done the test and a single fibre HBA (not the preferred connection for Violin admittedly) can only manage around 100k IOPS at 8K IO which is saturation point on bandwidth. This then causes queues at the server side which have the appearance of high storage latency.

      Although it seems on the surface that Violin would solve many issues without much thought, I've found that considerably more design and test time is required because the traditional bottlenecks are replaced with ones most IT folk are unaware of.

      1 0
  2. O RLY
    Headmaster

    I twitch when I see plurals such as "Proof of Concepts" or Attorney Generals".

    1 0
  3. Anonymous Coward
    Anonymous Coward

    seems like another decision in a long line of bad ones

    EMC OEM deal

    Hitachi OEM Deal

    Palm Acquisition/Sale

    EDS acquisition/write-off

    PolyServe acquisition

    Lefthand acquisition

    Ibrix acquisition

    3PAR acquisition

    Tying only 2 products from Violin to only BCS Server Sales (DL980 or HP-UX i2 products).

    Overall it does not seem like HP still knows where it wants to go or how to get there. At least they are focusing on doing something with Investor money now that they are focusing on their 3PAR purchase as their strategic direction. I am sure Hitachi has no issues with this at all....

    It is just a matter of time before Vertica becomes a huge revenue generation machine followed closely by Autonomy.....

    2 2
    1. Matt Bryant Silver badge
      Reply Icon
      Trollface

      Re: seems like another decision in a long line of bad ones

      "EMC OEM deal....." Everyone used to have to OEM EMC just as everyone had to OEM Brocade Silkworms back in the day, as EMC was the default choice for top-end enterprise UNIX. IBM learnt this to their cost. HP dumped EMC when they realised the Hitachi-based XP was much, much better than Symmetrix, etc.

      ".....Hitachi OEM Deal....." Which allowed HP to not only poach EMC customers but also work with those multi-vendor UNIX customers that wanted one storage device for UNIX, mainframe and Windows.

      "......Palm Acquisition/Sale....." Deabateable. I'm not sure how many patents it landed hp for the $1.2bn but it cost them less than the $12.5bn Google paid to buy Motorola's patents.

      ".....EDS acquisition/write-off...." IIRC, hp's consulting services are still making money.

      ".....PolyServe acquisition....." Que? That just sounds like sour grapes considering hp bought Polyserve from under Dell's and IBM's noses.

      ".......Lefthand acquisition...." Yes, because they haven't sold any P4000 or VSA systems, have they? Oh, wait - yes they have! AIUI, a lot of them displacing old NetApp NAS.

      ".....Ibrix acquisition...." <Yawn> See Polyserve above.

      "......3PAR acquisition....." Growing marketshare, making profits. You must be an ex-Sun director if you think those things are bad for business.

      ".....Tying only 2 products from Violin to only BCS Server Sales (DL980 or HP-UX i2 products)....." It looks like Violin was only a hold-over until they got a fully-SSD 3PAR sorted. One of the warnings with Violin is that no-one has bought them despite the hype. Violin has left it too late and it looks like the hype-bubble will burst before the IPO.

      1 1

      1. This post has been deleted by its author

      2. Anonymous Coward
        Reply Icon
        Anonymous Coward

        Re: seems like another decision in a long line of bad ones

        So eloquently you elaborate on all the ineptitude demonstrated by HP.

        Wonder why you stopped embellishing after violin, what about autonomy or Vertica.

        Lets not forget somemofmthe others.

        Apollo

        DEC (not HP per se)

        Tru64 and trucluster ( good enough for oracle RAC but not HP)

        Project Sage ( you might have to think on that one a bit)

        Convex

        With all of the of the capability HP has amassed - they should be doing more than growing market share for just 3par, Eds should be doing more than covering operating costs.

        You must still be an Hp employee who didn't take the buy-out when the stock price was above $14/share.

        1 1
        1. Matt Bryant Silver badge
          Reply Icon
          FAIL

          Re: Re: seems like another decision in a long line of bad ones

          ".....Wonder why you stopped embellishing after violin, what about autonomy or Vertica....." Well, I stopped showing you up after Violin because I have previously worked with all the other lines mentioned with several companies. I haven't had much to do with Autonomy or Vertica, but give it a few years and I'm pretty sure we'll be able to revisit the subject and prove you wrong again.

          ".....Apollo......" Apollo made workstations and were once the number one on their market in which hp had their own product, the 9000 workstations. But that number one position was long before hp bought them to integrate their tech into the hp 9000 workstations, IIRC they had fallen well-behind Sun in the workstation market by then (around about 1989?). The hp UNIX workstations then went on for another twenty-odd years before in turn being killed by the rise of more capable x64 workstations. How you translate that into some form of "disaster" would be interesting, but I suspect all you did was go look for a list of hp purchases and thought you could just list them unchallenged.

          "....DEC (not HP per se)..." A bit of list padding-out going on, I suspect. You really do need to go do a lot more reading before trying again. DEC were bought by Compaq, hp then later bought Compaq. D'uh!

          ".....Tru64 and trucluster ( good enough for oracle RAC but not HP)....." Tru64 was good enough for Oracle, but more customers wanted hp-ux and Serviceguard. The market killed Tru64, all hp did was give it the coup de grace. But I like the way you point out another case - just like Exadata - where Oracle didn't have the knowledge to make their own product.

          "....Project Sage...." Sorry, I'm not an hp internal, so if you have some FUD that's been handed you with an hp internal project name you'll just have to expand on it, thanks. Obviously, seeing as I can't Yahoogle anything for "hp project sage" I'd have to say it was so minor an issue it only featured on the radar of other vendor's FUD teams. Try again, little troll.

          "....Convex...." Convex had been trying to sell themselves to hp for years before hp finally did buy them up. The Convex machines were based on PA-RISC tech and were integrated into the hp range in the top-end V-class hp-ux servers which used to be quite popular in the enterprise. Again, how that translates to a "disaster" I'm not sure, maybe you want to actually supply some argument to go with your cobbled-together list? Or is that just asking waaaaaaay too much of you?

          ".....With all of the of the capability HP has amassed - they should be doing more than growing market share....." Fine, lets do some comparisons. How about IBM, another mega-corp which has made lots of acquisitions? Check the IBM figures for the last quarter, they're down. EMC? Flat-lining despite having the VMware money-printing machine. Sun? Already died, being zombified by Oracle.

          "....You must still be an Hp employee...." Whatever. Read some previous posts, I categorically state I am not an hp employee. But don't worry, I wouldn't accuse you of working for IBM or Oracle as you obviously just don't know enough to work in the industry as anything other than an XIV admin. I will however treat you to the taunt I previously only used for the terminal Sunshiner cases.

          /SP&L.

          0 1
          1. Anonymous Coward
            Reply Icon
            Anonymous Coward

            Re: seems like another decision in a long line of bad ones

            your funny - since you don't work for HP - why don't you stop supporting them so vehemently.

            yes I am privy to a lot internal information from all companies HP has acquired, as well as Oracle, the former Sun corporation and IBM as is my right as a 401K holder for many of the aforementioned companies.

            My argument is that for all the great companies and technology HP has acquired. It lacks the capability, direction, vision and conviction to do something with it.

            I worked for HP during several tours of duty - back when they actually had a right to the "What if" commercials.

            Regardless of IBMs own track record of devouring technology - they have maintained their shareholder value.

            And yes DEC was acquired by compaq and hp acquired compaq etc. but since you are a fan the register. I thought might find your way to the reference yourself but here is the link -

            http://www.theregister.co.uk/2004/12/02/hp_ends_tru64/

            Mellor and crew need to step up - us former corporate employees used to rely on the register to provide us update and timely roadmaps .... that seems to be dropping off lately....

            to re-iterate - the disaster is still the same, convex, apollo, tru-cluster, project sage (the foundation to the HP/ORacle Database machine), EMC, Hitachi, 3PAR, Violin, Lefthand, WebOs, etc. etc. no-one at HP is building solutions anymore. No-one is thinking outside the box and leading innovation. They have amassed a huge amount of ammunition. Just no plan, capability or common sense to build an arsenal to lead.

            http://www.youtube.com/watch?v=zNCrMEOqHpc

            0 2
            1. Matt Bryant Silver badge
              Reply Icon
              FAIL

              Re: Re: seems like another decision in a long line of bad ones

              "your funny - since you don't work for HP - why don't you stop supporting them so vehemently....." Two reasons - firstly, I do use and like a lot of their enterprise kit, which may not be perfect in every respect, but does do teh job very nicvely, thanks; secondly, I just can't stand whining trolls spreading FUD, and enjoy debunking what you amateur elmers post.

              "......yes I am privy to a lot internal information from all companies HP has acquired, as well as Oracle, the former Sun corporation...." Ah, no wonder you don't have a clue about tech and are so bitter - you bought Sun shares! HAHAHAHAHAHAAHAAAAA!

              /SP&L

              0 1
              1. Anonymous Coward
                Reply Icon
                Anonymous Coward

                Re: seems like another decision in a long line of bad ones

                You are still funny. <sigh>

                0 1
                1. Matt Bryant Silver badge
                  Reply Icon
                  FAIL

                  Re: Re: seems like another decision in a long line of bad ones

                  "You are still funny. <sigh>" Wow, what a comeback! Do I feel humbled by your wtt, eleoquence and outstanding display of technical knowledge. Well, actually I don't. Thankfully you've run out of rediculous FUD when challenged to provide a little meat to your arguments (what a surprise - NOT!). But, just for fun, please do explain why you think hp's bringing out an SSD-based 3PAR, which can be applied to more systems in their ranges AND competitors' systems, is somehow a "failure" compared to the OEMed Violin offering for a select few hp servers? Please, just to give us something more to laugh at.

                  /SP&L

                  0 1
  4. Anonymous Coward
    Anonymous Coward

    Still Funny -

    you are funny because all you can focus on is the things HP has done. Not what they are capable of doing or the capability of what they used to be able to do and vacuous absence of originality that permeates the company.

    FUD is about Fear Uncertainty and Doubt. I have not published any such statements. Having worked during the Lou, Carly, Mark and Leo years. I have had the pleasure of meeting and working with some of the smartest in the industry. From ol' days of test and measurement, to healthcare and the large systems group building MPE/HP0UX platforms.

    It is not FUD to state that HP is rudderless. it is self evident in how they operate. The divisions they sell-off. The companies/technologies they acquire.

    They lack direction, vision - hell conviction. This is a systemic issue with the shell that is left that is HP. As a customer you know how well and efficeintly they can take your order to how many proliants you would like to take deliver of this quarter or how many additional arrays/drives that you need to have installed at your secondary site.

    Ask them to solve your business issues? Ask them to help you innovate a new business process that helps your company take a strategic lead in your operation. With all the technology and all the capability HP is not innovating.

    3PAR and HDS have supported SSD devices for a while now. Issuing a press release that 3PAR/SSD is the direction is worthless. It was last year. As customer of HPs are going to state that you have not heard about installing SSDs in the existing 3PAR frames even the XP/P-Series from HDS? or even the storage blades and managing them with Lefthand software. Certainly not - you have already mentioned that you know how to google. So you have certainly had the opportunity to fact check anything your HP Sales team has been to update you on. Remember Agility...

    Can you sales person suggest that Violin Arrays with a DL380 Gen8 server is the best solution for your needs - or will you hear endlessly about what is on the truck,,,,

    Whether we like it or not the IT world is moving to more generalized cloud based service providers for back-office application delivery.

    With all the insight and capability that Mott brought to the greater HP ecosystem and the massive D/C consolidation story that he shlepped around the world to speak about - How is possible that no one, Not Mark Hurd, Not Ann Livermore, Not Martin Fink, Not Mike Nefkins/John Vestin (do I need to mention leo) and apparently Meg can figure out how to take the global services capability of EDS and their Tier1 Data Center Capability and deliver a service model that rivals Amazon et al in providing elastic compute/storage services.

    With MPE, OPENVMS, HP-UX, Tandem OS, seem like one or mabye three people at HP might understand multi-tenancy and being able to lock to OS services models to provide a secure operating platform.

    With their co-engineering with HDS set to expire/renew soon perhaps they can even continue on building highly-available disk based elastic storage farms.

    As for SAGE - again not FUD - but you will have to dig deep at HP to find anyone there who worked on it.

    It was project kicked off between HP and Oracle to develop a data warehouse platform that enabled greatly improved storage performance IO Bogged down systems such as Itanium - getting results for all those predicate chains requires a lot of IO/bandwidth to keep those caches full.

    There was a small group that did try to innovate - remember NeoView - right idea, wrong platform (possibly) definitely not the right leadership to pull it off.

    A appliance based approach to business intelligence will never work ( Exadata, Greenplum, Neteeza Teradata,...)

    It is not FUD to state the HP could not buy a great idea if PT Barnum tried to sell it to them.

    So yes as long you focus on the trees HP keeps talking about your posts to my statement are funny - because like HP you are focused on the wrong things.

    0 1
    1. Matt Bryant Silver badge
      Reply Icon
      FAIL

      Re: Still Funny -

      "OMFG, they haven't got a VISION!" Thanks, but you can take your vision and shove it. What you mean is "I am a shareholder and I want to hear some beguiling story about vapourware that will make my shares jump". I am a customer, I really couldn't give a damn about vapourware or "vision" as what I actually want are real products available NOW. Having "vision" and SFA else was what killed Sun. I am not an hp shareholder so I couldn't care less about boosting your stock price. For the last decade hp have actually done a very good job of listening to us customers and producing what us customers want. It went a bit off the rails with Leo, who tried to do the vision thing, but I don't see that as a reason to write them off. Oh, sorry, does that not help your pension fund? You must have me confused with someone that gives a monkey's.

      And why the heck would hp want to try going head-to-head with Amazon, Microsoft or any of the other big clouds? Hp would rather provide the hardware and software to sell to those cloud companies. Major, major fail.

      0 1
    2. Matt Bryant Silver badge
      Reply Icon
      FAIL

      Re: Still Funny - sorry, forgot to point out this bit of your stupidity

      ".....3PAR and HDS have supported SSD devices for a while now...." IRC, the current 3PARs are limited to 128 200GB SSDs per controller node pair, giving you about 4.4TB useable, which is not going to give you a very big all-Flash array. If hp do produce some new tech to stretch this to more SSDs then that would be very new and different. What, they didn't mention the SSD limit in your stock appraisal?

      0 1
  5. Anonymous Coward
    Anonymous Coward

    Still Funny -

    laughing my ass off - you make such wonderfully wrong assumptions about who I am, what my point of view is or what my interest may or may not be. On top of that you take everything literally perhaps you should step away from the keyboard and find a life...

    You clearly do not know anything about the products that you take enormous time to comment upon. You clearly lack the experience working with any of these products beyond your keyboard and what google can tell you.

    do you ever have the feeling that people are not laughing with you but at you...

    For someone who does not want a vision you certainly seem to espouse all rudderless direction that HP provides you.

    0 1
    1. Matt Bryant Silver badge
      Reply Icon
      FAIL

      Re: Still Funny -

      Yeah, whatever. I see you've run out FUD, just not out of hot air yet. I also see the technical points have you completely stumped - this is my surprised face, honest! You waded in with some assertion that hp has "failed" on a big list of purchases, then when challenged you fail to substantiate any of your assertions. And now you're just trying getting bitchy? Ooh, I'm so impressed - not! Come back with some actual meat next time, or don't bother, little troll.

      /SP&L

      0 1
This topic is closed for new posts.

Other stories you might like