back to article Cisco brags of industry-best hyperconverged performance

Cisco's HyperFlex hyperconverged products can support more than twice the number of VMs than other HCI systems while maintaining high performance, according to tests by ESG [PDF]. ESG testing used HCIBench, using Oracle's Vdbench tool, with the HCI products running virtual machines. It tells us the extensive testing "was …

  1. baspax

    My guess

    Vendor A - Nutanix

    Vendor B - Dell/EMC

    Vendor C - HPE (LeftHand VSA)

    See? Wasn't so hard.

    We are going to see the usual storage industry butthurt howling from all the usual suspects but Chris, you've been in this so long, you sure could have done better? How about some insight what a piece of crap VSAN really is. And how surprising it is that Nutanix isn't actually as bad as everyone thought. Don't get me wrong, Nutanix storage performance totally sucks if you try to do anything useful but the less than mediocre performance is probably good enough for all those suckers that run it at 10% utilization (like their large government customers).

    Let's be real for a moment. Hyperconverged was a stupid marketing fad championed by Nutanix to sell Software Defined Storage. Now that SDS is finally accepted we are seeing all the usual enterprise relevant features and criteria move front and center: performance, consistency & predictability, scalability, support, et al.

    The real story here is how Dell/EMC completely shat the bed with choosing VSAN as their enterprise SDS and how Pat Gelsinger overruled the old guard at EMC. And how Meg saw that circus and told everyone to hold her beer because she could screw it up even more by acquiring SimpliVity.

    SDS is moving front and center into the enterprise and all these morons do is acquire SMB software and somehow try to shoehorn it into the data center. Kudos to Cisco for trying to do the right thing here.

    1. Broooooose

      " .... all these morons do is acquire SMB software and somehow try to shoehorn it into the data center. Kudos to Cisco for trying to do the right thing here"

      Errrr, isn't that exactly what Cisco have done with SpringPath?

    2. Anonymous Coward
      Anonymous Coward

      Bitter much?

      Someone sounds like they never learned to play nice with others (and got side lined as a result).

  2. baspax

    Maybe, I was wondering the same. But it looks like Cisco stumbled (by sheer luck and coincidence most likely) upon a diamond in the rough when they evaluated SpringPath. Lord knows there was def no one who would have understood that scale-out SDS and fabric interconnects are a match made in heaven.

    What we do know however is that all these little startups lined up to get acquired by HP, Cisco, and Dell and for some reason Cisco (who were OEMing SimpliVity and to some extent VSAN) said no to all of them and instead picked up SpringPath in its embryonic phase. I honestly don't know how that happened. It looks like a cunning master stroke but I am at a complete loss as to WHO at Cisco is capable of recognizing and executing something like that.

    I guess we will never know or it will surface in a decade in some obscure story that only us old-timers will pay attention to but for now it looks like solid tech with quite some potential. Personally, I dismissed UCS when it launched years ago so I am a bit more cautious ruling out Cisco when it enters a new market, storage fumbles notwithstanding. Perhaps we are just giving them too much crap because of the Whiptail fiasco.

  3. Broooooose

    I think they have executed well, ie made the most from what they have. I don't think SpringPath was ever that exciting or revolutionary, but the ESG report serves them well. In truth, I've only ever seen HyperFlex in one account, and even then it was a test VDI environment for 100 users (an NHS CSU with budget spare at the end of the year). I come across a significant number of HCI opportunities every year and never see Cisco in them. It's nearly always Nutanix and increasingly VSAN (if you can call VSAN HCI?), as most accounts have a VMware vSphere implementation.

    I must say that I have to disagree with you about HCI being a stupid marketing fad to sell SDS. If that is your true opinion, you're in for a shock over the next few months and years. As I see it, Cisco's HCI implementation is fairly rudimentry, no shocks. It ticks the basic boxes but no more. Look at companies like HyperGrid, and to a lesser extent Nutanix. They see HCI hardware unpinning a much more ubiquitous private/hybrid cloud strategy, an applications/ services drive strategy. They want to compete against the AWS and Azure's of this world. Self service provisioning, automation and orchestration, in an on-premise environment to enables better control. They are already one step ahead of Cisco and the simplicity with which they have delivered their solutions puts them ahead of the likes of VMware/IBM too. I see HCI taking large chunks out the traditional server+storage market. Whether, the start ups are successful in their own right or just get swallowed up, this is the way the market is moving. However, I guess there will always be a group of people who pine for LUNS and yearn for Fibre Channel, and, for a while yet, they'll have plenty to kit to fill their boots with.

    1. Cloud, what..... Sorry... Um... - you just made that up.

      I kind of agree and I think HCI is a stupid marketing fad to sell SDS.

      I do agree with you that some of the HCI vendors have good API integration for a proper private cloud strategy, but really that comes down to automation and we could have done that years ago with scripts if we had the appetite. There are going to be customers that go HCI for the private cloud but I believe they are niche and won't drive much HCI demand when you consider the price premium for HCI.

      If you want proper automation, self service provisioning, charge back etc, do it propelly and use the public cloud because private cloud offering are imature.

      HCI will start to take more of the traditional market but I think a lot of that is down to marketing Hype a not any really technical reasons.

  4. Anonymous Coward
    Anonymous Coward

    NVCMe?

    Last paragraph - what is that? haven't seen the term before.

    1. tostaypuft

      Re: NVCMe?

      It seems like Google hasn't either

  5. Anonymous Coward
    Anonymous Coward

    Read the test here's where it's rigged...

    1. They tested a Hyperflex system with 10K capacity drives, against a system running 7200RPM SATA drives? I wonder why write latency was awful under steady state load on that platform.

    2. Vdbench to simulate SQL? Are they drunk? Can't afford benchmark factory, or SLOB?

    3. Why was a completely different model/ generation of SSD chosen for Vendor C? Based on Cisco Supply chain it looks like someone shifted from SAS to SATA here and to a several year old model.

    Missing also from this are the version numbers, firmware, drive make models, CPU's used, What the size/spec of the VM's were.

    If you want to see what a real methodical bench marking of HCI looks like Check out the Principal technology reports.

    Here's one for Nutanix.

    http://i.dell.com/sites/doccontent/shared-content/data-sheets/en/Documents/xc730xd-12-nutanix-acropoli.pdf

    Here are Exactly the same tests performed on vSAN

    http://www.principledtechnologies.com/Dell/PowerEdge_FX2_VMware_Virtual_SAN_0615_v2.pdf

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2019