VCE chooses Cisco ACI as SDN Strategy Instead of VMware NSX

Lede: I doubt that anyone bought a Vblock because it was a market leading networking strategy. I’ve been a supporter of VCE in the past but I’m doubtful that choosing Cisco ACI is good strategy for customers. My view is that most customers buy VCE Vblocks as a VMware platform. What VCE offers is a services package of fully supported storage, compute and networking that removes customer risk around VMware deployments while reselling EMC & Cisco hardware.

Note: I wrote this piece in September 2013 but I only just re-discovered it in my blogging queue.  I find it relevant six months later.

vce-sdn-strategy-game-on-gamble-595-opt.pngFor this reason, it is a big deal when VCE chooses to go forward with non-VMware software as announced by Trey Leyton here. This will be damaging to customers who chose the VMware lifestyle and want to continue with VMware NSX in the future and its tight integration with VMware vCloud.

I wrote a more comprehensive piece at Network Computing VCE Chooses Cisco SDN Over VMware NSX - this piece is an overspill of ideas and partially in response to some tiresome vitriol, again, from VCE employees and to explain my position more completely.

VCE have announced the decision ahead of the product launch and this means that customers can’t effectively evaluate the solution. Cisco is reasonably consistent at delivering new products late and pre-announcing details far too early. So even if ACI is fully announced in 2013 there is no reason to expect usable and reliable products before 2H2014.

VCE customers are VMware aligned and would be expecting to deploy NSX. VCE will have to work hard to convince customers that Cisco’s ACI strategy is a worthy product. We know very little about ACI yet and cannot determine it features compared to NSX for which there are demo versions and freely available labs online.

Cisco ACI will be unproven when compared to VMware NSX. NSX is an iteration of Nicira NSX which has been in deployment for 2 or 3 years at high-profile customer sites, it is reasonable to assume that this will translate into maturity and stability. Cisco ACI, by comparison, will be “version one” and will likely need some time to stabilise.

Cisco ACI best features are, probably, that it will be multi-vendor and multi-platform for hypervisors.  This is something that VCE business model doesn’t do. The VCE strategy is to limit choice and focus on a narrow range of products. For example, VCE does not support Hyper-V or OpenStack today while Nexus 1000V supports all of these platforms. Customers can choose to run those systems but there is no VCE value in doing so because VCE provides no support, design data or configuration advice. The VCE price premium is difficult to justify when you go outside the approved products. This is tough call for VCE since, like many other resellers, they don’t have enough resources or a business model to support both NSX and ACI. VCE’s value is limited to the professional services package that comes with the off the shelf hardware offered, again this measures VCE like any other reseller.

VCE choosing to sell ACI is of significant value to Cisco. I trust that VCE got something from Cisco in return. Admittedly, VMware has only a 10% stake in VCE so they are a minor partner.

The Vblock network of today is a trivial, non-core activity that exists at the fringe of the existing network strategy. Thus the Vblock purchasing decision receives little or no focus from network architects during the purchase cycle. A couple of Nexus 5K in a few racks rack of servers is not a significant network strategy change. By comparison, implementing Cisco ACI has much deeper ramifications on network strategy across the entire company. This likely to lead to impacts on buying cycles and purchasing decisions because the decision becomes orders of magnitude more complicated.

VCE is not a networking company and will need to build a new team with competency in SDN networking on virtualization systems. Not an easy task in the current financial climate but I’m sure there are plenty of people willing to step up to that challenge. Equally, engaging customers on Cisco ACI could be complicate the sales process.

Cisco ACI should offer stronger APIs for Networking and Server management. VCE has an infrastructure management platform and I consider Cisco ACI will offer better capabilities than NSX. In addition, I expect that ACI will extend to integrate the UCS Fabric Interconnect and UCS Servers in the future. Perhaps there is some value in this feature that I cannot see here and once the ACI roadmap is public we can make a better judgement on this issue.

The EtherealMind View

I wish VCE the best of luck in finding use cases and communicating the value proposition that will encourage customers to maintain the VCE relationship with Cisco’s ACI in the networking stack before the product is even announced or proven. VCE has a lot of talent and well capable of building strong marketing messages because their business continues to grow and be successful. It is not immediately obvious that this is the right decision for customers but it’s certainly a brave one. The oft-repeated and unofficial VCE motto of “Game On” suggests that selling is just a game. While I could wish they took their business more seriously at least VCE have found something to differentiate itself and put their “SDN foot forward” early and provides customers with some degree of certainty.

Customers will need to decide which SDN strategy suits their needs in the years ahead and now Vblock customers will have a harder time than most. I look forward to watching how this pans out.

Note: To the Cisco employees who don’t seem to comprehend that customers have their own positive and negative perspectives. This article isn’t specifically criticising Cisco or ACI or VCE. It’s simply my view as a customer. No need to jump on twitter or email and send me personal abuse.

  • returnofthemus

    In light of recent reports I think it’s fairly safe to say that this tripartite experiment is falling apart at the seams (see CRN: VMware Building ‘Project Mystic’ Converged Infrastructure Appliance For EMC).

    It really didn’t matter whose SDN arcitecture VCE chose to go with they never really had a value proposition, they just made the case for mainframe computing and in many cases you can buy a mainframe cheaper than a vBlock.

    Given that it takes them around 45-days to build and another 48-hours on-site to get up running, deploying a mainframe would probably be quicker too.

  • http://kendrickcoleman.com Kendrick Coleman

    Full Disclosure: My name is Kendrick Coleman and I’m a VCE Employee.

    Greg, I need to correct some of your statements and make some clarity of this situation.

    To clear up any misconceptions, VCE will always allow a customer to run NSX on Vblock Systems. We already have customers doing that today (some big customers, in fact). The only difference is that ACI will be a configurable option coming from VCE manufacturing, as well as completely covered by VCE seamless support. The flip side, if you want NSX then VMware PS will be involved and support for NSX will still be a single call through from VCE. Only difference, VCE does a “warm hand-off” to VMware. VMware will be on the hook for troubleshooting.

    VCE has always proclaimed to be a “Cisco networking” company. That isn’t going to change.

    Another completely untrue statement to clear up is, “For example, VCE does not support Hyper-V or OpenStack today while Nexus 1000V supports all of these platforms”.

    VCE allow customers to purchase “bare-metal Vblock Systems”. These bare metal systems come with everything completely pre-configured except a hypervisor. Support follows suit. VCE will cover a bare-metal Vblock System all the way up to the blade, then stop at the operating system. That operating system may be Hyper-V, RHEL, Fedora, Xen, etc. It doesn’t matter. VCE will still support the hardware.

    At the same time, OpenStack is compatible with VMware vCenter environments. So of course VCE supports OpenStack, VCE just won’t take a support call for the OpenStack component, only the VMware pieces. Have you noticed that VCE is actually a member and contributor to the OpenStack foundation? http://www.openstack.org/foundation/companies/profile/vce

    VCE is making strides of developing a platform that meshes with our customer’s vision and makes acquiring the technology easier. To quote a customer, “You know why I buy Vblocks? Because I’m lazy. VCE has the engineering efforts that will help guide me in a direction that will support my mission critical workloads and minimize risk to my business”

    Game On.

    • http://etherealmind.com Etherealmind

      I’ll attempt to summarise your response

      1. The value of VCE vBlock is that a given VMware/Cisco/EMC configuration is fully supported from a single company. Proven designs, performance guarantees and one throat to choke. That is a key VCE value proposition (and pretty cool BTW).

      2. Nothing prevents me from using Hyper-V, Linux KVM or any other operating system or software on a Vblock. The components are off the shelf systems from the respective Cisco and EMC divisions.

      3. However, if the customer chooses to do this, VCE will not offer support as an integrated partner. VCE will support the UCS, VNX etc but not the integration of the stack. The customer will need their own resources to cover the integration risk.

      My point is that VCE has chosen Cisco networking when most people choose a Vblock for VMware functions. The lack of direct NSX support doesn’t make sense to me. Perhaps there is something I’m not seeing here.

      Hopefully thats correct.

      PS: I don’t play games. I’m working to make my customers a success.

  • http://etherealmind.com Etherealmind

    The word “Game” is clear. If you want to indicate to customers that you a serious company then you shouldn’t play games with their business. Period.

  • returnofthemus

    Sorry to interject……..

    Apart from the fact that you can obtain a SAP HANA optimised system from any one of a dozen vendors now
    playing in this space, Oracle arguing that their platforms are best optimised for running their applications and Cisco offering ACI on FlexPods (or whatever they’re calling them nowdays), what is exactly is your USP?

    When these systems were announced I couldn’t help but chuckle, what you guys once coined as ‘reference architecture killers’ has transpired into being little more, albeit very expensive with limited piece parts, when did the market shift?

    Prior to the release of VCE Vision, Kent Erickson – Zenoss
    Product Strategist summed it up best imo, “converged
    in name, not in operation”, probably words he now regrets saying, so in essence an infrastructure bundle lacking cohesive management.

    In addition to VCE customers being lazy, I’d probably also imply extremely short-sighted.

    Please continue……….