A kerfuffle between AWS and VMware arose a few days ago in the wake of AWS releasing a new product: AWS Management Portal for vCenter.  What is AWS Management Portal for vCenter?  It’s a VMware vSphere vCenter plugin that allows AWS virtual machines to be managed from VMware vSphere vCenter.  Sound complicated?  Okay – to make it simple (which this plugin definitely is not in any way, shape, or form – see more on this at the bottom of this post) – this new product allows easier migration and management of VMware workloads into AWS.

VMware a few days after this announcement put up its own blog post titled “Don’t be fooled by import tools disguised as hybrid cloud management.”
  VMware noted as primary deficiencies that the AWS Management Portal for vCenter does not:

  • Allow resources to be allocated back into your VMware environment (in the backup world, we call this failback.)
  • Allows deployment to only one public cloud endpoint.

In short, VMware is warning people that the AWS Management Portal for vCenter is like the old roach motel product – data goes into AWS but it doesn’t come out.  In my favorite line in the VMware post, author Chris Wolf notes: “Don’t be fooled by basic management that is tactically useful today but can lead to increased lock-in down the road.”

I find it touching that VMware and Chris have discovered a new-found concern regarding vendor lock-in.  My guess is that this concern arises primarily from the fact that VMware understands that one of the major threats to its hegemony in virtualization is no longer just Microsoft Hyper-V, and no longer even container-based technology being popularized by Docker, but is also the hyperscale cloud vendors such as AWS, Azure, Google, HP’s Helion OpenStack cloud, Cisco’s ACI (Application Centric Infrastructure) OpenStack cloud, and others.  But assuming I’m wrong, I’m looking forward to the public pronouncement by VMware that their VCHS (VMware Hybrid Cloud Service) is a threat in terms of vendor lock-in as well. 🙂

A few words about what this plugin really is.  This is a product that has the following characteristics:

  • 1.4GB (that’s GIGABYTES) OVA download.
  • A single vCenter per AWS account.
  • Restricted to only vCenter 5.1 and later versions.
  • Limited to only IE10 and beyond browsers.
  • Appears to wrap-around the existing AWS “import API” which has very poor performance.
  • And…well…you get my point.  Not precisely a simple management or import product.

What does this have to do with backup?  Rather a lot in our opinion.  We believe that vendor lock-in is a major issue – whether it’s on-premise or in the cloud – whether it’s VMware or AWS or other vendors.  And we think that over the next few years backup and data protection will be revolutionized by the cloud – to an even greater degree than it’s been by virtualization.

What do you think?  Agree that VMware is being a bit of a hypocrite concerning its complaints that AWS isn’t open enough?  Think that AWS with its unwieldy management/import plugin is a wolf in sheep’s clothing?  Whatever you’re thinking, I’d love to hear from you.

Comments

  1. I think Unitrends is missing a significate opportunity. You should release the UEB as an AMI. That way, enterprises (like us) that have standardized on Unitrends, can leverage the same technology in their public cloud environment.

    Hopefully it won’t take as long as it took for Unitrends to support a non-proprietary cloud archive endpoint (not that I’m calling you guys hypocrites 😉 ).

  2. Chris: Thanks tremendously for writing. Let me ask you a question. What if we released UEB as an RHEL application – that you could run anywhere you run RHEL. Would that meet the need you articulate for AMI?

  3. Mark:
    Thanks for the super quick response. I think releasing the UEB as an application would definitely allow Unitrends to be deployed into any IaaS. IMO 100% a good move (let me know when it’s available, we’ll be happy to beta test). Though the UEB as an appliance is nice as the OS maintenance is abstracted as well.

    As far as Dropbox, I admit I used “non-proprietary” in haste. The point is that until the latest beta the UEB could only leverage one or two cloud endpoints. CloudHook (PhDVirtual ?) obviously changes that and we’re (im)patiently waiting on the beta.

  4. Chris: Thanks so much for the note. The Cloudhook technology is currently in beta in the 7.5 release (went out last week.) Write to beta@unitrends.com (or simply write to me at my last name and the Unitrends domain) and we’ll get you set up on it.

    On the UEB as an application, thanks for the feedback – and stay tuned – we’re currently scoping this out – initial plans are to release before end of year.

Comments are closed.