Throwing Rocks at OpenStack: Open Source as a Distro or a Service?

As open source shifts from a distribution model to "as a service," some mistake the transition for a weakness in the technology.

I’ve been fortunate to spend nearly my entire career working at companies deeply engaged in open source — at Red Hat for 11 years, then Puppet and now as the vice president of engineering at Rackspace. In that time, I’ve been involved in several technology shifts that have changed the industry. It’s been fascinating to watch them evolve — especially when folks are throwing rocks.

And rocks are thrown. Every. Single. Time.

Today, as we gear up for the latest OpenStack Summit, this one in Boston May 8-11, OpenStack finds itself in the crosshairs, as the industry shifts yet again, from offering open source infrastructure using the traditional distribution model to “as a service.”

OpenStack is following the path of most successful open source software efforts before it, with explosive early growth around functionality and community, along with natural cycles of innovation and failure. As usage steadily increases and the core matures, we’re at the point where the fundamentals work well. A lot of efforts are now focused on ‘nice-to-have’ features, while innovation has the latitude for adjacent technologies and what’s next — a great position for any project to be in.

Early commercial entities dove into early OpenStack efforts headfirst, partly to help build great technology, yes, but equally motivated to become the next Red Hat — to become the dominant distribution player in the market. Now, however, we’re seeing companies remember an important distinction: open source is a development model, not a business model.

That’s led to rock throwing in the form of cries that “OpenStack is dead,” (a sentiment my colleague Scott Crenshaw rebuts in this post), reminding me of the early growth years of Linux. There were an extraordinary number of challenges back then: competition (there were more than 100 Linux distributions), patent trolls, conflicts with historical purchasing and license models — and getting enough razors to shave all the yaks.

Two decades ago, a software distribution model was necessary for the success of open source. First, open source had no reliable consumption or business model. Second, hardware vendors needed the equivalent of an operating system vendor to narrow the combinatorial hellscape that could have been. Equally important for developers and application providers, distribution allowed users to focus on their central concern — running their own businesses. Ultimately, the software world ended up consolidating around Red Hat’s various flavors of commercial and free Linux.

One of the early benefits of distributions centered around which components were included and kept “updated” for users. But over time, the packages in the leading commercial distribution became stale; concern about indemnification evaporated, open source licensing became better understood, and consumers became more savvy. As open source matured, so did practices around delivery, meaning organizations could go off to get the components they needed directly to source and consume; many were formally packaged by groups external to commercial distributions.

All the while, open source proved itself reliable, innovative and more importantly, a commodity. Today it’s everywhere: in phones, cars, televisions, data centers. It’s making sure my smoker controls the temperature for the brisket cooking as I write this. We’re at a place where open source is a given, and lucky us, OpenStack is one of the largest open source communities in the world.

And while OpenStack is helping shape the next generation of open source, it’s also showing us the element that needs to be completely rebooted: the distribution model. It’s simply no longer sufficient. Spoiler alert: it’s about operability — and what can be built on top.

Ultimately, utility cloud providers have exposed how difficult it is to properly operate data centers — and reminded all of us that the ability to expertly operate infrastructure is what really fuels the consumption of open source infrastructure.

Luckily, today’s IT consumers are increasingly embracing the “as a service” model of consuming infrastructure. I say luckily because while running OpenStack continues to get easier, at large scale it — like all cloud infrastructure — is still difficult to manage and run. The reality is, infrastructure isn’t a product. It’s a living thing, requiring expert care and feeding. It takes deep thinking and deliberative operations engineering. Every business benefits from expert operations, yet it remains elusive, especially in-house.

But expert operations is something Rackspace has been for almost two decades — longer than anyone else. While other vendors are finally rushing to create managed service offerings, we already have more than a billion hours of OpenStack server experience under our belts, and we run clouds five times larger than any other vendor. Whatever the challenge, chances are we’ve already encountered — and solved — it.

Concretely, that means we’ve come up with a set of patterns, tools, processes and customer engagement that start companies with the right size deployment and then scale up — and that means markedly better economics. It also means flexibility; your timeline to “cloudify” your applications, so to speak, can be controlled by you. If you want to go fast, we can help you go faster. If you want to be more deliberative, we can help you do that, too.

But most critically, it means expertise. More and more companies are doing the math on what it takes to attract the talent, and continually invest in the expertise, to manage all these disparate services. Similar to not rolling a Linux distribution, they’re finding that it’s simply not cost effective to develop and maintain expertise in the nuances of operating complex cloud infrastructure. Better to spend that investment building more effective applications on top of a living managed infrastructure service.

OpenStack-as-a-service insulates you from that, allowing you to focus precious resources where they can do the most good — creating value for your business.

The distribution model is dying, and that’s okay. It served its purpose, and is now being replaced by something better, something that works for today’s enterprise — and tomorrow’s. They need cloud to work economically for all their workloads, not just those that run well in public clouds.

More than two dozen Rackspace OpenStack experts will be on hand at the Summit next week, you can find their speaking sessions here, or swing by Booth B20 and have your questions answered directly.

Can’t make it to Boston? No problem. Sign up for a no strings attached free consultation to speak with one of our experts about your business objectives. We’ll strategize with you on the solutions you need to achieve them. SIGN UP NOW.

Brian Stein serves as Vice President of Global Engineering at Rackspace, and was recently appointed to the OpenStack Foundation Board of Directors. Brian has more than 18 years of experience in open source development, user experience, quality engineering and assurance, release engineering and project management. Before Rackspace, he served as senior vice president of Engineering at Puppet Labs, furthering the configuration-as-code to an enterprise-class platform, which helped form the foundation of the modern DevOps movement. Prior to Puppet Labs, Brian spent more than a decade at Red Hat, where he served as vice president of engineering, building and leading a worldwide team delivering Red Hat's family of cloud products, including virtualization management, OpenStack, cloud orchestration and management, Linux and middleware management through core virtualization enablement. Brian was a key leader in Red Hat's transformation to cloud infrastructure, and created and led one of the largest OpenStack engineering teams in the industry. He holds a Master's Degree in Computer Science from The State University of New York Binghamton in Vestal, NY.

LEAVE A REPLY

Please enter your comment!
Please enter your name here