Forging a path from telco to cloud with Microsoft Azure
This blog was co-authored by Angel Shimelish, Sr. Product Marketing Manager, Azure for Operators.
Across industries, organizations are venturing deeper into their digital transformation journeys every day, and the telco industry is no exception. Angel Shimelish recently sat down with Ryan van Wyk, Microsoft’s Partner Software Engineering Manager, who joined Microsoft as part of our acquisition of AT&T’s carrier-grade Network Cloud platform technology. Ryan is helping to lead the transition of mobile networks to an open, disaggregated, and software-based model, leveraging Azure for operators as the platform of choice. Angel and Ryan covered a range of topics about his personal journey from AT&T to Microsoft, his focus on innovation, and how he sees the future of cloud transformation technology impacting the telco industry.
Cloud transformation for telco
Angel Shimelish: Thanks for meeting with me to discuss your experience in bringing telcos to the cloud. Before we delve into your professional background, let’s discuss your interests outside of work. Tell me a bit about your favorite hobby.
Ryan van Wyk: I am a sailor, I started sailing as Sea Scout while growing up in South Africa and was fortunate enough while living in Chicago to spend nights and weekends sailing on Lake Michigan. Nowadays, I look forward to my family’s annual sailing adventure, where we charter a sailboat and explore a new part of the world from the water. What I love about sailing is you are constantly learning, improving your skills, and it requires a fair amount of planning, and as my wife Danielle will attest, I don’t like to sit still.
Angel Shimelish: In one of our past conversations, you mentioned you were really excited about this opportunity to continue to lead your team. Can you share your perspective?
Ryan van Wyk: Personally, being able to lead the high-performance telco cloud engineering team that was assembled over seven years, is amazing because one, from a human connection perspective, we’ve got a deep level of trust across our internal team, and two, from a software perspective, we can take what we have built to new heights. By bringing the team to Azure for operators, we have been able to concentrate the knowledge, experience, and learnings over several years building Software Defined Networks (SDN) at AT&T and now at Microsoft. This is very powerful because when assembling a new team, you often spend years building that team, the trust, figuring out where your strengths and weaknesses are, and it takes time to develop a shared set of learnings, and as a result, you may lose valuable time in trial and error. We’ve made our mistakes and are bringing these learnings forward into Azure for operators, and that’s extremely powerful in terms of our ability to deliver a proven telco-grade product to operators. It’s not just about the software, it’s about the people who build the software.
Angel Shimelish: While at the OpenStack Summit in Barcelona in 2016, you identified a novel solution by inverting Azure Kubernetes Service and OpenStack. Can you explain this idea that was the genesis of the Airship Open Source project and how you achieved this problem-solving ability?
Ryan van Wyk: In 2016 the industry was dealing with the life cycle management challenges of OpenStack, which were magnified when using OpenStack as a NFVi to run operator network functions. My team reviewed a project called TripleO (OpenStack on OpenStack) to potentially solve some of these challenges, but as we discussed it, we realized that many of the same pitfalls would still exist, but now, at a different layer of the stack. The discussion within my team led to a question: Why not hold the software we used to run our cloud infrastructure to the same modern software and cloud practices we were pushing our VNF partners to embrace; portability, microservices, and immutability? Everyone was discussing running Kubernetes on top of OpenStack, but could we do the inverse and containerize OpenStack itself, and run it on top of Kubernetes? If modern software was being continuously integrated, delivered, and deployed using technologies like Docker containers and Kubernetes, could we not do the same for our platform? Wouldn’t this solve not only our NFVi delivery speed and lifecycle problems, but thinking ahead, could it also prepare us for Cloud Native Network Functions (CNFs) even though this use case was several years out? As we discussed the idea further, and the benefits of a declarative approach to software delivery, it dawned on me that we had something here, something innovative.
To clarify, this is not an innovation I own, it was a team effort. Even though it happened while sitting around a table at an OpenStack conference, the reality is the idea formed as the result of many discussions over several months. It would have been easy to get overwhelmed with the challenges that we faced with our massive scale OpenStack deployment, fight the fires, and blame the state of the technology, but my team and I needed to own the decisions we had made, and the results. Only by doing this were we able to acknowledge what the root problems were that needed solving, and then formulate the design principles for the new platform, Network Cloud. So, I learned a long time ago to listen to your team, constantly be challenging the status quo, ask lots of questions, and be willing to acknowledge failure.
Angel Shimelish: Often, you reference the need for predictability when defining a solution that is easy to understand and replicate across various stages of a project or solution. How did you come to define this quality as being important in your work?
Ryan van Wyk: Achieving predictability has been key to our success with Network Cloud after the trial and error deploying prior versions of the NFVi around the world. It has allowed us to build trust with operation teams, this has led to more intelligent risk-taking, which enabled the rapid scale-up of the platform. For example, if we know that the test we run in our engineering labs will yield the same results in production, then the velocity at which we can introduce new features into production can increase because we can trust the production experience will be what we have observed in earlier stages of delivery. To this end, we shift all our testing left into earlier stages of delivery, and as a rule ensure the same tests run in development, test, and production acceptance. Additionally embracing a principle of everything delivered to an environment is containerized has helped to ensure that our Network Cloud deployments are truly immutable which is another key element of predictable delivery.
Angel Shimelish: You’ve noted that you believe that in order to be successful, a team “needs to constantly assess its performance, pay attention to what others are doing, and listen to the feedback of users”. How did you adopt this philosophy of openness, and how do you see it influencing your new work with Microsoft?
Ryan van Wyk: This was something learned through working in open source communities. The beauty of open source is you get to speak engineer-to-engineer about the challenges you’re facing and share what you have learned. This can be a humbling experience as you are exposed to how engineering teams all over the world are working, and you quickly realize you don’t know everything, or that there are multiple ways to solve a problem. Also, when trying to build something in the community, feedback on your work is continuous, comes from a different perspective, and is not influenced potentially by internal corporate considerations. I believe this approach is going to be essential for our success at building a best-in-class solution in Azure for operators. To do this we are working closely with a number of Azure engineering teams to build on their great work and are partnering closely with AT&T and other Operators on the direction our platform is taking.
Angel Shimelish: In closing, can you share your vision for telco digital transformation and how the work you and your team are building are going to help realize the potential of digital transformation?
Ryan van Wyk: Microsoft can supply operators with cloud computing options that meet the customer wherever those capabilities are needed: at the enterprise edge, the network edge, the network core, or in the cloud. The various form factors, optimized to support the location where they are deployed, are supported by the Azure platform—providing a virtual machine and container services common management framework, DevOps support, and security control.
Richer network experiences through Azure
By combining these flexible computing options and exposing these services through the Azure platform, we can combine them with other Azure capabilities to bring the power of AI and automation to the delivery of network services. These capabilities, in concert with our partnerships with OSS and BSS providers, allow us to help operators to streamline and simplify operations, create new services to monetize the network, and gain greater insights into customer behavior in the network. We are also able to combine the network with other capabilities in Microsoft to create a deeper, richer network experiences, with the flexibility that operators need.