20 Comments

Summary:

As the latest release of the open-source cloud infrastructure debuted, controversy swirled anew. Will OpenStack kill third-party PaaSes or vice versa?

openstacklogo

As the OpenStack Havana release went live last week, controversy over the viability of the open source cloud continued to swirl, this time thanks to a set of dueling blog posts.

On Saturday, Simon Wardley posted a blog post expressing his grave doubts that the OpenStack cloud can reach its potential to be the basis of a competitive public cloud ecosystem.

“I don’t see a benevolent dictator or a flourishing market of competitive providers with easy switching between them at scale. I do see a collective prisoner dilemma, a refocus on private cloud (a future niche) and capitulation in the public space with Amazon and Google gaining speed.”

Wardley, a researcher at CSC’s Leading Edge Forum, recommended that folks who want to see a booming open-source cloud ecosystem turn instead to Cloud Foundry, the open-source PaaS backed by Pivotal, the EMC/VMware spinoff.

“I’ve got high hopes for Cloud Foundry to create a competitive market at the platform space around open source. I would wish that companies rather than trying to introduce alternative approaches which blur the market in order to support their dying past business models would just adopt. I expect them to however argue that differentiation on a commodity is key and the same old nonsense be repeated again.”

The OpenStack gang will likely have a vivid response to this. Wardley’s post seems at least partly a reaction to an earlier Mirantis blog. There Alex Friedland, co-founder of Mirantis, a member of the OpenStack Foundation, posited that the addition of richer feature sets layered atop OpenStack foundations  – like the metering and orchestration– in Havana will kill third-party PaaSes. He cited Cloud Foundry and Red Hat OpenShift specifically as potential victims. Yikes.

That caused quite the outburst in the twitterverse, especially given that Red Hat is a primary contributor to OpenStack and the game plan of Cloud Foundry is to provide a cloud-agnostic PaaS layer atop AWS, OpenStack, whatever.

And, the whole notion of adding PaaSy capabilities to IaaS is hardly new. Amazon Web Services has been doing this for years.

Still, please use comments to register your thoughts on this controversy. Will a feature-rich OpenStack negate the need for PaaSes? Or is this a specious argument? Serious question.

Everyone can probably agree that discussions at the OpenStack Summit next month will be lively and one key concern — the ability of different vendors’ OpenStack clouds to interoperate — will be one topic of conversation.

Other cloud news from around the interwebz:

From GigaOM: VMware makes desktop virtualization move in buying Desktone.

From Seeking Alpha: Amazon announces new Cloudfront CDN features.

From Data Center Dynamics VMware brings cloud services to Europe.

  1. I think there will always be a place for special PaaS software, however they will have to be more tightly integrated with OpenStack in the future. Out of Cloud Foundry and OpenShift I think OpenShift will come out ahead due to being sponsored by Red Hat that is also a big contributor to OpenStack.

    I just don’t see where VMware will have relevancy in the future, they’re like Solaris but for virtualization and OpenStack is Linux (we all know who’s alive and who’s on a dwindling fall to irrelevance…)

    I have to recognize that VMware has a huge market share today, but how will VMware be able to justify the huge license fees they’re taking for their software in the future when OpenStack is good enough or better for the 99% of all companies?

    Share
    1. This comment is such a #facepalm … oh well… it’s Sunday…

      Share
      1. Well, Hyper-V is very relevant to Microsoft shops and also significantly less pricey compared to the other “enterprise” alternative. But where’s the value in VMware when the hypervisor market is already commodity and OpenStack is being the new standard for rolling out private clouds. Is VMware’s management software really that much better that I’m willing to pay that much extra for it?

        Also VMware is only in the OpenStack group by force (Nicira). If they could, they’d give it the kiss of death in a heartbeat.

        Share
        1. #facepalm #2

          Share
        2. CloudUser1, hiding behind an anonymous nickname makes me feel we are in the eighties. Please fix that and come out with your affiliation.

          Share
    2. Trevor Roberts Jr. Sunday, October 20, 2013

      When making the VMware\OpenStack comparisons, it’s important to clarify which VMware components you are referring to.

      OpenStack is fully capable of managing VMware’s hypervisor via Nova to deploy Cloud workloads. You can take the integration for a spin by downloading VMware’s VOVA Project.

      Customers who enjoy the benefits of VMware’s hypervisor platform do not face an either-or dilemma if they want to deploy an OpenStack Cloud. They get the benefits of the OpenStack framework coupled with the reliability and performance of the vSphere platform.

      VMware acknowledged, with the acquisition of DynamicOps, that they want to help customers manage their multi-hypervisor environments. I don’t see their membership in the OpenStack Foundation as a compulsion but rather as another logical step in helping their customers manage diversified IaaS infrastructure.

      Share
    3. I work for a large financial company and we will not buy into OpenShift simple because RedHat plays the same games as VMWare does when it comes to licensing. We would rather go with OpenStack and CloudFoundry from two different vendors rather than OpenShift+OpenStack from a single vendor. Avoiding vendor lockin is the goal of all these different ecosystems and it is nice to see that although each vendor has an agenda to make money of these products they are still striving to maintain compatibility. Our goal is to sideline any vendor who chooses to create lock-in through some promised customization which experience has shown ends up being buggy and costing us more in the long run. We’d rather let the community do the work to evolve and build a robust IaaS/PaaS eco-system like we have seen in the Java community – a.k.a. Spring. By the way Pivotal sponsors Spring and IBM fostered the eclipse community, both very successful open source communities.. and yes they both sponsor CloudFoundry :-)

      Share
      1. i hear this a lot re. Red Hat — that it’s a very pricey “open source” solution. ifyou’d like to discuss further @mevansam plse email me or DM me on twitter @gigabarb.

        Share
  2. Jesse Proudman Sunday, October 20, 2013

    RedHat is an OpenStack Contributor and supports OpenShift.

    IBM and HP both are active OpenStack contributors are support Cloud Foundry (HP via ActiveState).

    Looking forward to finishing up my blog post on this topic.

    Share
    1. agreed. VMware, one of the parents of Pivotal/Cloud Foundry is also an OpenStack player by virtue of its Nicira ownership.

      This is why i want more discussion on this topic. Plse post a link to your post here @jesse.

      Share
      1. Jesse Proudman Sunday, October 20, 2013

        Hi Barb,

        My thoughts are posted here:
        https://www.bluebox.net/insight/blog-article/openstack-cloud-foundry-so-happy-together

        Happy Sunday!

        - Jesse

        Share
  3. James Watters Sunday, October 20, 2013

    The vital and scandalous background to all of this is that Mirantis was fired from a large Cloud Foundry customer recently.

    Remember that on the internet anyone can publish a blog….

    Share
  4. Of course Mirantis has it’s own agenda, but so does every other company. Pure PaaS companies will try to show that they’re still relevant in the long run. Cloud Foundry will most likely have a great future ahead of it. The question is, will Pivotal make enough money on it to stay afloat when OpenStack is gaining PaaS features?

    Red Hat can always bundle and tightly integrate OpenShift and OpenStack into their own solution to add extra value, but can Pivotal really do that with VMware as their owner?

    Share
  5. James Watters Sunday, October 20, 2013

    Lots of this is really about Mirantis making a childish attack on other Openstack distributions that have better market traction and CF integration like Piston, which was selected as the default build server for the Cloud Foundry integration, as covered by Barb:

    http://gigaom.com/2013/08/22/old-pals-cloud-foundry-and-piston-cloud-collaborate-on-paasiaas-integration/

    IBM is also a major Cloud Foundry sponsor, and 100% focused on Cloud Foundry on Openstack. Mirantis is a very minor player trying to make noise with a blog–and, hats off to them, doing well with that!

    Share
  6. I’ve been around a long time but not long enough to remember if there was similar gamesmanship claims/counterclaims among the various Linux backers when Linux was finding its footing — but i suspect it happened then too.

    Share
  7. The ‘controversy’ seems a bit irrelevant as innovations like Docker are already turning deployment & operations focused PaaS solutions into commodities.

    At Nirmata, we believe that next-gen application platforms will be focused on specific architectures allowing them to address the entire application life-cycle. This will allow for true disruption rather than providing incremental benefits to traditional application architectures.

    Share
  8. Jeff Schneider Monday, October 21, 2013

    On the date that the Mirantis blog was posted – the author stated that there was more interest in the platform services that are ‘broad but shallow’ (general purpose orchestration & configuration) over ‘narrow but deep’ (constrained/focused PaaS offerings).

    This point isn’t really up for debate – the data lives in Github, Indeed or just by asking the guy down the hall. Today, more people use CloudFormation, Puppet, Chef, Salt, Ansible or other solutions that enable ‘platform automation’. They don’t claim to be PaaS – but they do claim to offer some (not all) of the benefits. Conversely, they do purport to solve a broader set of use cases than our currently offered by PaaS solutions.

    Organizations will use a combination of general purpose automation tools side-by-side with specific PaaS solutions. It’s not an ‘either-or’ scenario. I’m not sure why the folks at Pivotal got so upset over this reality. Their parent company (VMware/EMC) is a significant investor in PuppetLabs. One would think that the they would focus more on creating a combined solution of topology orchestration and last-mile configuration rather than bad mouthing the inevitable approach! How odd.

    Share
  9. Joseph Glanville Monday, October 21, 2013

    The problem is they are all missing the point. The public cloud debate should die. It’s obvious Amazon and GCE are going to be the #1 and #2 respectively. (there is never a #3)

    Openstack however is doing AMAZING in the private cloud space. All of the Openstack vendors are posting big wins.

    Even if we allow for the public cloud being a $10B (it’s not that big) business that is still realistically 2% of IT spending (roughly $2T in 2013).

    So why do we care so much about public cloud? Is there really much evidence to support people moving off on-prem to public clouds? Or are they just deploying new applications that don’t need to be on-prem and they never would have had before? (aka mobile and web apps)

    From where I am sitting I am seeing all the BUA workloads stay sitting on metal in DCs owned an operated on premise and that is not going to change in the next 6 years.

    Share
    1. Not sure I agree with you 100% there. I think Rackspace will be nr2 or nr3 in public clouds. But I agree with you, OpenStack will be an amazing private cloud platform to build on.

      Share
  10. I would encourage everyone here to read this HEAT blueprint: https://wiki.openstack.org/wiki/Heat/Software-Configuration-Provider. It may not be exactly PaaS, but it’s definitely moving in that direction and if I was a PaaS product manager, I’d be scratching my head reading this…

    My take is that Red Hat figured out that a) it won’t be CF with OpenShift; b) OpenShift can’t be merged into OpenStack. So what they are after is building an OpenStack native PaaS from within, bottom-up. And if OpenStack will have a native PaaS-like service integrated with core infra projects, it is likely to work much better than a 3-rd party, platform agnostic PaaS running on top.

    Doesn’t mean that PaaS is dead, or CF is dead. Means that there is a clear threat to PaaSes looking to run on top of OpenStack coming from within the OpenStack community. And those who simply choose to ignore and discount it are, IMHO, in denial.

    Share

Comments have been disabled for this post