Building the future with open ecosystems
A conversation with Rajesh Gadiyar,
Vice President and CTO, Network Platforms Group, Intel
As part of our interview series exploring the four dimensions of openness driving telecom innovation, Stefan Kindt, Solutions lead at Nokia’s Cross Portfolio Marketing spoke to Rajesh Gadiyar, Vice President and CTO of Intel’s Network Platforms Group.
In a wide-ranging interview, Rajesh and Stefan discussed the role of open source and open APIs in driving the next generation of network and edge applications, which will increasingly be delivered on a cloud-native model. An edited version of their conversation is below.
Vice President and CTO, Network Platforms Group, Intel
Solutions Lead, Cross Portfolio Marketing,
STEFAN KINDT: Rajesh, tell us about your role at Intel, and within the telecom industry more generally.
RAJESH GADIYAR: I've been at Intel for about 20 years, primarily focused on network applications. In the last few years I've been working on network function virtualization and software defined networking, both of which have seen really good progress in the industry.
More recently, I've also been working on 5G. The past year has been a tough one for us all, but we’ve made some really good progress in 5G RAN and cloud native 5G SA Core in spite of the pandemic. And one of my other favorite areas of work is cloud networking, which is the next phase in the network transformation.
SK: And when we talk about openness in the telecom industry, what does that term mean to you?
RG: To me it’s all about community-driven innovation, with a goal to drive faster transformation in the marketplace. That requires an open mindset, open platforms, open interfaces, and the ability for everyone to create value together. The beauty of an open approach is that we all win. We win together, we win big, and our customers and end users benefit as well.
SK: Where do you see the biggest impact from openness? Is it about time to market, sharing ideas, cost reduction – or is it a combination of those?
RG: It's all of those things. Openness is extremely important as the network continues to go through a transformation that’s accelerating with the rollout of 5G.
When we have the ability to take open standard specifications, define an open architecture and create open platforms that everyone can innovate with, it means we can all get solutions to market much quicker. That’s how openness allows us all to win through collaboration. But it also allows us to deliver innovative solutions for our customers and drive the industry forwards much faster.
SK: Can you share any examples with us, based on your own experience, where openness has accelerated innovation?
RG: A good example is the open source Data Plane Development Kit. It’s changed the face of networking by providing a software stack that allows high-speed networking – hundreds of gigabits per second today – on a standard high volume server platform.
Going forward, massive cloud-scale automation is going to be critical for networks and CSPs as well. That's another place where open community-driven innovation has been super-helpful in terms of progressing the industry and achieving cloud scale for network applications.
SK: Openness hasn’t always been part of the DNA in the telecom industry. Have you seen a tension between the traditional standards-based approach and the open source architecture approach of the cloud evolution?
RG: When you have a 100-plus year old, heavily regulated, foundational utility service like telecommunications, there’s a lot of legacy and ingrained ways of doing things. Regulation and standards sometimes form an inertial force that makes it extremely difficult to achieve agility – especially compared to a new infrastructure like cloud computing, where there's virtually no legacy, and the old rules don't apply.
Collaborative software development, for example, is still in its infancy. But I am extremely proud of the progress we've made over the last 10 years or so in the disaggregation of hardware and software. That’s been instrumental to driving a faster innovation cycle and has led to some great examples of network virtualization across the board.
Intel and Nokia have done great work together with the wireless packet core, for example. As we look towards standalone 5G and the user plane functions, I can't imagine getting to the true potential of 5G without having done that work of network virtualization and hardware-software disaggregation. It's allowing us to move faster and realize the end-to-end benefits of things like network slicing.
SK: When it comes to innovation, do you see a conflict between open communities and more limited co-creation and joint innovation partnerships?
RG: No, I don't see a conflict there. There may be a piece of IP – an algorithm, say, or an accelerator – that provides a significant differentiation.
There are instances when a business needs to keep the IP closed. To me that's not in conflict with the openness tenets. There are many instances when our customers want to build something proprietary with us – a proprietary ASIC, for example. As long as it still fits the paradigm of openness around the support for open standards, open interfaces, I think it's fine.
SK: And what about the balance between open source and proprietary software? Can a company like Intel participate in the open-source community and also offer commercial products to customers?
RG: For me they're two sides of the same coin. Open source is important because it allows the community to come together and innovate faster. There’s also a good co-development model.
At the same time, many CSPs don't have the process in-house to be able to take open source and deploy it in a commercial production environment. There’s also always a need to extend the open source project into something tailored to a certain use case or deployment model. So it’s important to have a commercial ecosystem that provides software that’s continuously maintained, deployable, and supported.
For example, at Intel, we participate in many open source communities, including Linux Foundation Networking and the Cloud Native Compute Foundation. They’re very important forums to drive innovation and bring new services to market faster. At the same time, our Intel® Network Builders ecosystem is exclusively focused on driving robust, commercial, supported software solutions.
SK: Can you talk about the innovation you’ve driven with open APIs? And how important is it to build an ecosystem to innovate on top of them?
RG: I’ll answer in terms of the new paradigm of cloud native, where I’m trying to drive three things.
First, I'm trying to drive openness around the hardware platforms, and the disaggregation of hardware and software. Applications in the cloud aren’t built for a particular platform or a particular hardware, so the ability to build applications to run anywhere becomes very important.
The second thing is automation at scale. We've been working very closely with communities like Kubernetes and ONAP on Enhanced Platform Awareness capabilities. Kubernetes and ONAP are the window through which we can expose the hardware capabilities in this disaggregated and hardware abstraction model.
Thirdly, it is about composability of applications and services – the next generation of IoT and network applications will be built on cloud principles, and that’s all about composability, scale and flexibility. We’re increasingly talking about a microservices-based deployment model with APIs such as REST and gRPC, the ability to take a service or an application and break it into component microservices that can be containerized and deployed anywhere in the network is powerful.
SK: Of the four dimensions of openness, does any one stand out for you in terms of driving innovation?
RG: They’re all important. Open standards and communities like 3GPP and ETSI are very important, because that's how we define standards and specifications in a collaborative way. Then you take those specifications into open architectures where everyone can innovate. O-RAN is a good example. We’re beginning to see it as foundational in terms of how we open up the radio access network and drive innovation, which will be fundamental to driving the true benefit of 5G.
And then comes open source, which allows you to take the open standard specification and the open architecture, and you convert it into an implementation with community-driven innovative development. This is where projects like ONAP and DPDK play a critical role.
Open interfaces have a key role as well. eCPRI is a good example of an open interface that’s allowed the disaggregation of radio units from the distributed unit. And then finally that allows for an open ecosystem to come together to move the industry forward.
SK: Lastly, on a scale from 1 to 10, how important is openness for driving innovation – especially as we move deeper into 5G, IoT and Industry 4.0?
RG: I’d rate it 9 or 10, because openness has already helped us innovate much faster. 5G wouldn’t have been deployed without the progress we’d made with NFV and the disaggregation of hardware and software. So openness is a critical element of how we’ve driven faster innovation.
Looking to the next phase of cloud-native – towards edge and distributed computing – and applications that will have a high dependency on AI and automation. Openness will be critical to driving innovation over the next 10 or 15 years. And that’s something I find super exciting.
SK: Thank you very much Rajesh for your time and your insights.