OpenStack has spent years languishing as an "anybody but AWS" community standard. While the community of cloud vendors may not love Amazon, the community of cloud developers and enterprise customers clearly does—to the tune of billions of dollars every year.
And yet, there may just be a little hope in OpenStack land. On Wednesday, OpenStack vendor Mirantis announced a collaboration with Google and Intel to rewrite key portions of OpenStack's infrastructure as Docker containers managed by Kubernetes. For the historically-sluggish OpenStack project, this is a dramatic move. And a promising one.
By collaborating with Google and Intel, perhaps the feisty startup can finally accelerate innovation in OpenStack and give customers a private cloud platform that can better keep up with AWS. I recently spoke to Boris Renski, co-founder and CMO of Mirantis, to find out more.
TechRepublic: What is the big news with Mirantis, Google, and OpenStack that you are announcing?
Renski: We are containerizing all of the OpenStack services using Docker's container format and are standardizing on Kubernetes as the container orchestration fabric. What that means is that OpenStack users will be able to manage OpenStack in the same way that Google manages all of their internal services. Another added benefit is that users will get a single fabric for containers and VMs. We're collaborating with Google and Intel on this effort.
The problem with OpenStack has been that it was historically built by people that have little experience running large scale, distributed systems. That is, once the cloud is installed, how do you make sure it doesn't break, how do you patch it, how do you upgrade? All questions that the OpenStack community hasn't put sufficient emphasis on.
Because of that, enterprises have been failing to run OpenStack in a stable fashion. They can install it, yes. But then it stops working after a month and nobody has a clue how to fix it.
Google runs all their internal systems, including Google Cloud, in the very specific way that they perfected over the course of many years. Kubernetes is a tool they released that allows one to run a large scale, distributed system using this "Google way." We are taking this exact "Google way" of running software and modernizing OpenStack architecture to adhere to it, making OpenStack more stable and easier to manage.
TechRepublic: How will this change how customers run their OpenStack clouds?
Renski: Instead of doing a big release twice a year and requiring that customers do a forklift update, we'll be able to deliver a true infrastructure-as-code experience for the customers. With this new model, we'll be able to move away from the traditional software paradigm of incremental OpenStack releases that we ship to customers every six months. We'll be driving the customers to adopt a continuous integration, continuous delivery model for their infrastructure. We'll be continuously shipping updated containers into customer's staging environments, which they can promote to production. Mirantis OpenStack will become a single, continuously rolling release.
TechRepublic: What's in it for Intel and Google?
Renski: Intel cares because they want to foster on-premises clouds everywhere, the core of their Cloud for All initiative. The whole idea of Google pushing their technology out in this fashion greatly facilitates that. Intel has a lot of resources and know-how at the bare metal level, as well as monitoring the physical infrastructure. They can make enormous contributions to OpenStack by showing us, and Google, innovations in advance.
TechRepublic: There is a big market share battle today around container orchestration. And while I've written about Kubernetes early community lead, isn't it too early to be picking a winner?
Renski: As we see the space mature, there will be standards evolving. With containers themselves, Docker today is the de facto standard. In the container orchestration space, Docker Swarm, Kubernetes, and Mesos are still battling. But for us, as insiders, it's evident that Kubernetes will be the clear leader with others playing a supporting role. Mesos will be around—it's a good old technology that works well for big data—but Kubernetes has the mindshare.
In 2011, when OpenStack was battling with CloudStack and Eucalyptus, we made a conscious bet to standardize on OpenStack and we were right. Now, we are making the same bet on Kubernetes when it comes to container management.
TechRepublic: Does this move help Google compete in the cloud wars for more private cloud market share?
Renski: We think that the broad theme in our industry is that the public cloud model, as it matures, is starting to shift away from the idea that one cloud dominates all use cases. Public cloud vendors need a path to serve customers on premises. Microsoft is pushing Azure Stack, and there are rumors that AWS is looking to launch an on-premises offering. Our collaboration with Google will help their foray into the on-premises market through OpenStack.
Less than 5% of workloads today run on public clouds. That means there is 95% of on-premises infrastructure spend still up for grabs.
Google is going after that market by taking technologies they've innovated on, like Kubernetes, and giving them to cloud developers and operators who will say, "this is the coolest thing ever!" They want to make sure there's mindshare around their stuff—dominating the private cloud. They did this with Android. They didn't want everyone to run Apple iOS and have Apple be the gateway to all mobile. By open sourcing Android, Google moved to front and center in mobile.
With Kubernetes, they're doing the same thing with containers and cloud infrastructure. OpenStack is primarily on-premises computing. It's a dominant open source fabric for on-premises infrastructure. That's the whole point to why Google is supporting this. The next frontier to winning the public cloud wars between Google, Microsoft, and Amazon is capturing the on-premises cloud mindshare.
TechRepublic: OpenStack is open source with a community and a well-established governance model. Can you, Google, and Intel just make this happen? Where does OpenStack's community love fit in?
Renski: Let me be clear. All of this work is going to be done with open source software in upstream OpenStack. It will involve work across multiple projects. There will be a lot of work around Fuel, OpenStack's most popular lifecycle management project, which we initially pioneered. There will be work in Kolla—the OpenStack project for containerizing OpenStack services, which we're also very involved with.
The way open source works in general—and OpenStack in particular—is that it's not so much about seeking consensus. There are a wealth of great ideas. But, if you take initiative and do good work, the best code prevails. We've socialized this broadly—some folks agree, some disagree. But generally, there's a consensus and folks are stepping up. This is us taking the initiative to make a great idea happen.
- Why Kubernetes could be crowned king of container management (TechRepublic)
- Hey ops teams, developers want control of the data center (TechRepublic)
- OpenStack is overkill for Docker (TechRepublic)
- Why Kubernetes may be a bigger threat to Amazon than Google's cloud (TechRepublic)
- Docker and Mesos: Like peanut butter and jelly (TechRepublic)
Matt Asay is a veteran technology columnist who has written for CNET, ReadWrite, and other tech media. Asay has also held a variety of executive roles with leading mobile and big data software companies.