On the Path to Open Infrastructure: A Conversation with Shane Wang
Earlier this year, Shane Wang was re-elected as an individual board director of the OpenStack Foundation, and is now serving in his fifth year. As an engineering manager in Intel’s Cloud and Networking team for fifteen years, with a focus on OpenStack over the last seven, Shane talks about his aspirations as a Board member, OpenStack in China, and the role of the OpenStack Board as OpenStack evolves to Open Infrastructure.
Working on OpenStack for over seven years, you’ve seen the project grow and evolve. What do you think have been some of the most pivotal milestones along the way?
There have been four key milestones that really stand out in my memory. Early in OpenStack’s history, there were a few projects that targeted the cloud infrastructure layer. Of these projects, OpenStack seemed to take the lead in popularity around 2012, becoming the de-facto standard for the infrastructure layer.
Initially, OpenStack was used for public clouds. The second milestone came around 2015 when we began to focus on its enterprise readiness, working on features like live migration and rolling upgrades, and its use shifted to private clouds. Since then, more and more businesses have adopted OpenStack as a private cloud solution.
Then we began to focus on ensuring OpenStack could address network function virtualization (NFV) for companies like China Mobile and AT&T. We’ve seen many developers from telco carriers join the OpenStack community and use OpenStack to support their NFV applications. We’ve also seen the emergence of many other open source projects, like OPNFV and ONAP, primarily based on OpenStack.
Today, I think we’re in the midst of a fourth transition—an expansion to open infrastructure where we’re targeting emerging user scenarios, like artificial intelligence (AI) cloud and edge computing.
It has been exciting to see the significant adoption of open source—and OpenStack—in China. Can you give us your insights?
More and more Chinese companies are adopting OpenStack. In China, there are more users of OpenStack than contributors to OpenStack. Users, like state-owned enterprises, are interested in adopting OpenStack, but don’t have the resources to contribute to the code. Through initiatives like the Open Source Hackathon series, we’re working to encourage developers from China to contribute to OpenStack, and we’re hoping to attract more users as contributors.
In your tenure as an OpenStack individual board director, which contributions are you proudest of?
I’ve really enjoyed helping increase awareness and knowledge about OpenStack in China, which in turn, has helped boost contributions to the project and accelerate its adoption in China. And one of my dreams has been to bring an OpenStack (Open Infrastructure) Summit to China, so I’m really looking forward to welcoming the community to China Mainland later this year!
As OpenStack evolves to Open Infrastructure, can you tell us what the term ‘Open Infrastructure’ mean to you?
Collaboration. Initially, OpenStack aimed to solve the problem of the infrastructure layer. However, we’re seeing more and more user scenarios emerging—software-defined networking (SDN) and SDN controllers, NFV, software-defined storage (SDS), edge computing, AI cloud. OpenStack cannot meet all of the requirements demanded by these workloads. We have Kata Containers to address the needs of secure containers, and we have StarlingX to tackle the challenges of edge computing. We need more and more projects—and collaboration between projects—to create solutions for different use cases.
As OpenStack transitions to Open Infrastructure to address workloads of the future, how can the OpenStack Board help ease this transformation?
I think the Board can act as a bridge between the OpenStack Foundation and the individual members in the community. On one hand, the Board should listen to the community members and OpenStack users—what they’re doing, what they want to do and what their pain points are—to raise their needs and requirements to the Foundation. On the other hand, we need to pass down new technologies from the Foundation to the individual members.
For example, when we started to work on StarlingX, there was no single solution to address the challenges of edge computing. When StarlingX became an incubation project, we needed to communicate to the individual members and users that StarlingX could help solve the problems and meet the requirements of edge workloads. If the community feels like the Board listens to and advocates for them, they will be more forthcoming with their ideas, and more and more individuals will get involved in the community.
How can the Board help incubate and embrace a greater number of open source projects under the Open Infrastructure umbrella, and help these projects collaborate with the OpenStack community?
The simple answer is that the Board can vote to incubate, or graduate, projects. But it’s more than just a vote—we need guidance in the form of working groups, and opportunities to come together. In the case of edge computing, the Board created an Edge Computing Working Group and provided opportunities for people to unite to raise requirements, share user scenarios, surface questions, and brainstorm solutions—all to solve problems.
What’s your favorite/most important OpenStack debate?
There’s been discussion around containers vs virtual machines (VMs), and different orchestration options. We have airplanes, cars and trains, in the same way that we have containers, virtual machines and bare metal. None of these can replace the other; all of these will co-exist. We need all of these, depending on the use case, or workload. If you ask telco carriers, many of them will tell you that containers cannot satisfy the computation performance needs or requirements that their businesses demand. Meanwhile, online shopping sites gravitate to containers; one of the largest online retailers in China boasts the largest Kubernetes cluster in the world. Clearly, these businesses are different and each requires a different approach. Some projects are simply more mature than others. OpenStack is simply more mature than Kubernetes. If your family has a new baby, all the attention will be focused on the new baby because the new baby needs more care than older siblings! In the future, I suspect we will have new technologies to satisfy new scenarios that are neither containers nor VMs.
It’s so vital to continue to welcome newcomers into the community, and inspire them to contribute. What suggestions would you give them as they consider joining the community and getting involved?
OpenStack has done a great job welcoming newcomers. If there’s one barrier for newcomers to overcome, I think it’s the language barrier. This is the biggest problem for folks from China to join the OpenStack community. In global meetings and discussions, community members from China will remain very quiet. This is not because they don’t have any thoughts; it’s because it’s hard for them to communicate their own ideas in English. Perhaps the hosts or chairs of these meetings can provide such individuals with more opportunities to speak up in these meetings.
What are you enjoying most at this week’s Summit?
I've attended so many OpenStack (or Open Infrastructure) Summits. This time around, I'm really enjoying seeing new and different things—new technologies and new user stories, like edge computing and AI cloud.