10 Comments

Summary:

Contributor James Urquhart takes issue with recent calls for OpenStack to build its own platform as a service. Citing the commercial success of VMware and Amazon Web Services, he claims the real benefit of any platform is its community.

The OpenStack summit in Hong Kong this week promises to be a pivotal moment in the history of the open source cloud platform. While the project continues to mature, the Folsom release was a major milestone when it came to code stability, and the Havana release promises to push the bar even higher with improved testing and a growing community of contributors.

Not all is sunshine and lollipops in the OpenStack world, however, and there is one recent controversy that concerns me deeply. Recent blog posts by Mirantis CEO Alex Freeman, and a proposed OpenStack incubator project around application lifecycle management (ALM), indicate there is a small but vocal community of contributors to OpenStack that do not understand the value of a consuming service ecosystem for OpenStack.

“Consuming service ecosystem”?

Because I may have accidentially just invented a new buzzword trying to explain my point, let me start by explaining the term. OpenStack is a platform, in every sense of the word. It’s a set of code that forms a foundation from which a (hopefully) rich ecosystem of solutions to a wide variety of computing problems can form. It’s the soil from which a new marketplace can grow.

Unfortunately, in my opinion, all of the focus to date in OpenStack has been on the delivery of OpenStack itself, not in building a rich selection of solutions that consume OpenStack. Hey, it makes sense that a variety of system and platform companies, both established and newly created, would want to package, install and support customers who want to run OpenStack core services. There is a lot of money to be made in delivering platforms to customers.

However, the value of those platforms (at least in business) is dependent on the platform being a part of business solutions. While there are certainly IT departments that see “private cloud” as a business problem to solve, those projects won’t be successful unless there are either new revenue streams created on the platform, or significant cost reductions in existing business use of IT.

So, when I say “consuming service ecosystem,” I mean the ecosystem of customer and partner software solutions that consume OpenStack (via APIs or other means) to deliver their value.

AWS and VMware as examples

Two great examples of this are AWS and VMware. In both cases, you might think the most valuable thing they do is deliver virtual machines as a service. But if you talk to developers that rely on AWS, or operators that rely on VMware, you’ll hear that what makes those solutions valuable to their buyers: that they can actually solve problems using a combination of that core compute service and value-added capabilities provided by the respective vendor and partner (or community) ecosystems.

AWS has its own services, such as RDS and CloudFront, and VMware has a huge library of management and operational tools built to support its virtualization and cloud capabilities. So, each platform’s buyers don’t have to piece together solutions to complex problems with custom code and bailing wire — they can often find commercial (or open source), tested solutions that eliminate the vast majority of work.

The end result is that each ecosystem attracts more customers, which attracts more partners, which creates more solutions, which — in turn — attracts more customers.

Just a fraction of the huge OpenStack community.

Just a fraction of the huge OpenStack community.

Where OpenStack sends the wrong signals

OpenStack, on the other hand, has a small group of community members determined to prove that nobody — and I mean nobody — does open source cloud as well as they do. The Mirantis post that kicked off the controversy about PaaS and OpenStack made clear that projects such as Heat (a deployment automation service) and the newly proposed Solum (the aforementioned ALM service) are throwing down the gauntlet to those who would deign to try and add value for developers outside of the OpenStack project itself.

God forbid you might try to do so across multiple cloud services and platforms.

Let me be clea that I have a horse in this race, as Dell Multi-cloud Manager does indeed provide management for many cloud platforms, including OpenStack. But the point here isn’t that any one segment is being attacked. The point is that OpenStack should not chase away any project — even the established PaaS platforms — that might actually drive customers to the platform.

The signal sent here is horrible. Instead of saying, “Hey, PaaS platforms, we don’t want or need you” (or “don’t want to want or need you”), the entire OpenStack community should be hellbent on saying, “We love anyone who makes OpenStack more valuable, and encourage others to develop for the platform.”

What OpenStack should do

I think the Solum project is ill advised, but here’s what I’d love to see: a subset of the community who makes it their mission to reach out and liason with Cloud Foundry, OpenShift, Jenkins, and any other open source PaaS and ALM project worthy of attention. These liasons would not only make sure that the projects were advocated for within the OpenStack community, but help drive resolutions to technical issues, project release alignment and co-promotion opportunities.

I’d also love to see the OpenStack Foundation actively recruit software companies to build software that takes advantage of OpenStack, even if it is also built to run on other cloud services and platforms. Grow the consuming cloud ecosystem, and help promote opportunities to make money by consuming OpenStack services, rather than simply packaging and delivering OpenStack directly. Heck, even encourage competition within the OpenStack ecosystem.

While I, unfortunately, couldn’t make the trip to Hong Kong this year, I am genuinely looking forward to seeing the incredible progress OpenStack has made in the latest release, and hearing more about how it will get even more powerful in the future. I just hope that future is friendly to the world where software projects and vendors can benefit OpenStack while benefiting from it, rather a world that is just dismissive of key potential partners.

Thoughts? Disagree? Let me know in the comments below, or reach out to me on Twitter, where I am @jamesurquhart.

Feature image courtesy of Shutterstock user Istvan Csak.

  1. What’s the real problem here. Dell just needs to show that their PaaS + OpenStack solution is truly better than that of OpenStack and Solum PaaS. It sounds more like he doesn’t like the competition and is afraid the products he represent won’t measure up to the new competition?

    What I think Dell needs to do is bundle their Hardware with their OpenStack solution and PaaS solution. Then make is easy to consume and not require loads of consulting hours to get running :-)

    Share
  2. James – I was as dismayed as you were by Alex’s comments however I wouldn’t conflate this spat with the solum project. We are involved as are Red Hat OpenShift team and we are far from wedded to OpenStack or *Stack for that matter.

    Share
  3. James – I was as dismayed as you were by Alex’s comments however I wouldn’t conflate this spat with the solum project. We are involved with this as are the Red Hat OpenShift team and we are far from wedded to OpenStack or *Stack for that matter. There is a broad church and lively discussion. Of course it would be great to have CloudFoundry guys join the discussion.

    Share
  4. Jesse Proudman Monday, November 4, 2013

    Hi James,

    You nailed it and this resonates with my post “OpenStack Has One Job, Do It Well:” https://www.bluebox.net/insight/blog-article/openstack-has-one-job-iaas

    At Blue Box, we’re avid supporters of enabling external technologies that grow the core OpenStack technology suite. We believe one of the greatest benefits of OpenStack is the fact that it’s open. Writing applications and tooling that can run without modification on Public Clouds, Hosted Private Clouds and On-Premise Private Clouds will help the OpenStack ecosystem immensely. But OpenStack needs to make sure it’s empowering developers to do just that.

    – Jesse Proudman
    @blueboxjesse

    Share
  5. James, there seems to be confusion in defining who’s community here. Some of the existing players don’t consider CloudFoundry/ OpenShift to be not, while many of us do. I agree with you that reaching out to existing open source project makes sense. What I don’t understand though, why RedHat didn’t make any effort to bring in OpenShift while it is one of the partners of Solum.

    IMHV, it is a natural progression to add more layers (what people try to define as ‘core’ will keep expanding) and it needs to. Like how Linux Kernel is to an extent.

    However, I don’t agree with the total premise that OpenStack as a whole doesn’t understand the ecosystem just because some claimed so.

    Share
  6. I agree 100% about OpenStack not understanding about ecosystem.

    I attempted to build a service connecting to OpenStack but ran into a missing feature. When I tried to explain the use case, no one understood the idea that all instances of OpenStack should support the same feature so external services could connect to it.

    Share
  7. Facts:

    1) Solum is not aligned or coordinated with Alex Freeman’s opinion blog. They were not coordinated and are not related. The close timing was unfortunate.

    2) Solum is not proposed for OpenStack incubation. It is OpenStack Related, meaning that it is designed to work with OpenStack, and there are no plans to become part of OpenStack’s core software suite. This was outlined in our FAQ upon announcement.

    3) Solum is a collaboration among several companies interested in integrating with OpenStack Clouds who are interested in Solum’s vision.

    4) Solum is not part of any strategy of the OpenStack Foundation. Rackspace coordinated Solum as a collaboration for an open development effort to address key gaps in OpenStack that many PaaS software/service vendors have agreed upon.

    5) Nobody affiliated with Solum has said anything like “Hey, PaaS platforms, we don’t want or need you”. In fact, most of the contributors are in the PaaS business.

    Solum is about adding additional capabilities in a project that takes advantage of new capabilities in OpenStack, including Docker’s Nova driver, and Heat. The PaaS ecosystem can benefit from an API that surfaces these capabilities in a way that enables applications to be portable among public and private clouds.

    Regards,

    Adrian Otto

    Share
  8. Yeah, I don’t really understand what the fuss is about. What danger does Solum pose to CloudFoundry or any other PaaS? It’s like saying that Ubuntu poses a danger to RedHat because they work on competing server management pieces. Who cares? Let the best solution win.

    I find it a bit comical that a certain small contingent of self-styled cloud pundits have taken such umbrage with the thought that some individuals want to work on a new PaaS system. Maybe these individuals just aren’t interested in working on the other codebases or contributing to those projects for any number of reasons. There’s nothing wrong with that, and it doesn’t hurt any competing PaaS project.

    Best,
    -jay

    Share
  9. I replied earlier (but it appears to have been deleted), agreeing with Adrian’s points and his last paragraph in particular. I’ve expanded on ActiveState’s feelings on Solum in a blog post at: http://www.activestate.com/blog/2013/11/why-would-activestate-join-openstacks-project-solum

    Share
  10. Sorry I don’t agree with lot being said here. You lost credibility when you said you have a product that overlaps with Solum and that you don’t want yet another competitor in the fray. Sounds like products like RightScale have eaten your lunch? Oh well.

    Share

Comments have been disabled for this post