Clicky

Examining Marvell’s Key Role in Advancing O-RAN Innovation – Futurum Tech Webcast Interview Series
by Ron Westfall | February 2, 2022

In this episode of the Futurum Tech Webcast, Interview Series, we tackle the topic of driving O-RAN innovation across the 5G ecosystem and how Marvell’s O-RAN platform enables base station OEM partners to fulfill the demanding performance and latency requirements of 5G network while also being positioned to innovate their offerings as deployments and standards evolve. My guest today is Joel Brand, Senior Director Product Marketing at Marvell Technologies, a semiconductor company. Joel is a new guest on this show, and deftly shares his insights and perspective on the direction of the open-source O-RAN market segment and its pivotal role across the entire 5G ecosystem.

For the foundation of our discussion, we addressed why Open RAN architectures are being developed to provide operators with a choice of infrastructure supplier options for RAN components and increased flexibility. For our perspective, Open RAN implementations are moving beyond lab trials and greenfield deployments to mainstream mobile and private networks. Today, Marvell’s O-RAN platform provides a purpose-built, cloud native, highly programmable solution to address capacity, power, cost, and time-to-market challenges presented by current architectures that rely heavily on general-purpose processors.

Our conversation focused on the following:

  • Why O-RAN is so important to the future of 5G and why the open interface approach of O-RAN is so critical to attaining the top-priority outcomes identified by operations such as spurring ecosystem-wide 5G innovation as well as improving applications by optimal mixing and matching of solutions from suppliers.
  • A description of the Marvell O-RAN platform solution in reference to silicon, software, and hardware reference designs and how it differs from existing alternatives.
  • The rationale behind the O-RAN disaggregation of RU, DU, and CU functions and how Marvell’s solutions map across the architecture.
  • We explored why general-purpose processors are sub-optimal in handling the majority of 5G physical layer/Layer 1 (L1) functions and if the same limitations apply to O-RAN architectures.
  • How the Marvell O-RAN Fusion-O product line fulfills the infrastructure processing demands of L1 in O-RAN networks.
  • How Marvell’s solution can avoid trade-offs in features and performance while supporting the open scalability of vRAN technology across cloud environments.

You can watch the video of our conversation here (and subscribe to our YouTube channel while you’re there):

Or stream it here on your favorite podcast platform:

Don’t Miss An Episode – Subscribe Below:

 

Disclaimer: The Futurum Tech Webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors and we do not ask that you treat us as such.

Other insights from Futurum Research:

MWC 2021: Marvell OCTEON 10 DPU Launch Injects Innovation Boost to DPU Technology

Marvell Industry Analyst Day 2021: New AtlasOne Chip Readies 5G RAN Optical Fronthaul for 5G Ecosystem

Marvell Octeon DPU Family Boards the Burgeoning Evenstar Initiative

Transcript:

Ron Westfall: Hello, everyone. And welcome to this episode of Futurum Tech Webcast. And today we’re very pleased to have Joel Brand of Marvell Technologies join us as we look into examining Marvell’s key role in advancing O-RAN innovation across the 5G ecosystem. And before we get into it, I like to provide the reminder that today’s conversation is for educational purposes only, and is not to be construed as investment advice in any way whatsoever. And so with that, Joel, how are you today? How’s it going?

Joel Brand: I’m good. It’s good to be here. Thanks for having me.

Ron Westfall: You bet. Very pleased. And this is an opportunity. Tell us more about yourself and your role at Marvell Technologies.

Joel Brand: I’m looking after the wireless business of the processor business unit at Marvell. I’ve been involved in the wireless industry since the early days when CDMA and GSM were looking it out for world dominance, seems like about 100 years old ago or something like that. And I’ve been fortunate enough to go through the different generations of wireless technologies, and it’s been pleasure.

Ron Westfall: Yes. I have seen the same evolution of our industry. And so, yes, you’re definitely the right person to talk about O-RAN. And let’s start with that 0-RAN technology itself. And why it is so important to the future of 5G, that is, in other words, why is the open interface approach of 0-RAN so critical to attaining the top priority outcomes that operators emphasize. And this includes spurring ecosystem-wide 5G innovation and developing applications through the mixing and matching of solutions from suppliers. What’s your perspective?

Joel Brand: It is critical… It’s utmost important the open interfaces. And it’s not specific to 5G. It’s more general to the telecom and the wireless ecosystem at large. The operators are desperately needing innovation. They need the ability to replicate what has been successfully built on the internet. If you look at the TCP/IP networking industry, if you look at the software industry, it’s all built around open APIs, development tools, protocol stacks that are very well layered, even open-source communities that have flourished. And the telecom ecosystem has not had the opportunity to leverage and enjoy that. When you built the environment in such a bite-size pieces, a lot of newcomers, new companies can come in because you essentially lower the barrier of entry for new innovative companies. And you allow the ecosystem to innovate faster where not every company needs to build everything that Nokia has built over the last 200 years. They can build a small piece of it and contribute that to the ecosystem. And that’s what will spur the evolution and the innovation in the new telecom network once we’re able to institutionalize all the O-RAN interfaces out there.

Ron Westfall: Yeah, I think that’s very important. Clearly, 5G in itself is important, and it also must interwork and coexist with 4G technologies, cloud technologies, and so forth. And so, yes, this is, I think, very important to know as we explore why O-RAN technology is so important. And I think this really does give us the level set to look more into the Marvell portfolio itself. And that is, can you describe the Marvell O-RAN platform solution in reference to the silicon software and hardware reference designs, and how does the Marvell approach differ from alternatives that are out there in the market?

Joel Brand: Yeah, I appreciate the question. So, Marvell is very unique in the sense that this is the only merchant silicon available today for 5G radio access network processing. The alternative to that is to use a custom ASIC, but in terms of merchant silicon, something that is available off-the-shelf today, if you want to develop your own protocol stack, particularly the physical layer for 5G, this is the only available solution and Marvell has differentiated it on that.

On top of that, we obviously built an extensive software capabilities to allow customers to use the capabilities of the silicon in the most efficient way to integrate with, again, standout off-the-shelf environments that are widely available on the internet. We’ve built around it reference designs as you pointed out. And we’ve built around it a vRAN, Virtual Radio Access Network solution by taking the silicon and putting it on a PCI board. And that allows our customer to start enjoying the benefits of bringing the cloud in the RAN, Radio Access Network closer together and starting to leverage commercial off-the-shelf server hardware in order to run the 5G RAN network from possibly lower-cost hardware.

Ron Westfall: Yeah. And that’s a very valuable insight. I see the industry definitely warming up more to the approach, and I think it also goes as the opportunity to drill down more into what makes O-RAN work, what makes it so special. And that is… Can you talk about more about the rationale behind O-RAN disaggregation of the RU, DU and CU capabilities? And even equally important, how does Marvell solution map across this architecture?

Joel Brand: Yeah. We have more acronyms in this industry than we can count. RU, DU… So the concept is actually very, very simple. The acronyms make it sound complicated but the idea is to find the right balance between how capable the transport network, the network that connects your core network to the towers in the field. The balance between that and the complexity of the equipment on the tower. In general, what you want is to have a transport network that doesn’t need to transport too much, that the requirements are to transport less, to transport it slower, because the more you are asking to transport more, and the more you are asking for lower latency transport, the more expensive that transport becomes. You cannot use maybe millimeter-wave backhaul but instead, you have to use fiber when you put very difficult requirements on the transport.

On the other hand, you’re trying to simplify the equipment on the tower. The more complicated that is, the more maintenance it requires, the more expensive it is, the larger it is. And the idea is, the more we can simplify it, we need more transport because then we will do more the processing in the backend. The more we complicate the equipment on the tower, we would reduce the complexities on the transport. And that the balance is what the split of RU, DU, CU is designed to address. From a Marvell perspective, we are providing solutions for the RU where a lot of the Lo-fi [inaudible] and beamforming is done for massive MIMO radios. We provide solutions for the DU where most of the physical layer of the protocol stack is processed. And we provide solutions for the CU where a lot of TCP/IP networking and encryption and security functions are offered. That’s our OCTEON product line and on the radio. And the DU it’s our OCTEON Fusion product line.

Ron Westfall: And that’s very insightful. This is a great explanation, Joel. What we see is the operators prioritizing the ability to use disaggregation to align the network to support specific use cases to really enable them to meet the specific needs of customers and naturally monetize it in a more flexible way. And I think this provides the opportunity to look at the market from a different perspective. And that is, can we look at the general-purpose processes that are deployed today and perhaps explain it why general-purpose processors are suboptimal in handling the majority of 5G physical layer one functions. And also do these same limitations apply to the 0-RAN architectures themselves?

Joel Brand: Yeah. General processors are important and we all use them on our laptops and many other places but the 5G physical layer processing cannot be done in a general-purpose processor. And it doesn’t really matter if this is a vRAN architecture, an O-RAN architecture, a traditional radio access network, a distributed architecture, it doesn’t really matter. A general-purpose processor was designed for general purpose. It’s like using a general-purpose car. My Honda is not ideal for transporting a king-size bed, for that I better use a truck or if I want to race, I better use a race car. You want something specialized when you are trying to perform a specialized function. And the 5G physical layer is a very specialized function. We just need to envision the complexity of that.

We are trying to counter the effects of the air interface between the tower and the mobile phone. And users are holding the phones sometimes in their pocket; sometimes next to their ear; sometimes they are driving. There are many users that each tower needs to handle and there are trees and rain and all kind of other things that are interfering with the transmission between the tower and the mobile user. And we need to counter that. We’ve built a very, very complicated math in order to be able to counter the effects of that air interface. And that’s what the physical layer is all about. And because of that complexity, a dedicated silicon is required in order to handle the physical layer. A general purpose processor simply cannot do that. And again that’s true for vRAN, O-RAN or any other types of radio access network.

Ron Westfall: Yeah. I agree. That’s I think, a well-established trend, something that is simply essential that we need specialized processors for infrastructure processing and that we simply have moved beyond using general processors certainly for these applications and capabilities. And I think this also leads into revisiting the Marvell OCTEON Fusion product line. And that is what makes a product line so different. And so purpose design for fulfilling these infrastructure processing demands, especially for layer one requirements and any O-RAN network out there.

Joel Brand: Yeah, it’s a delicate balance. We talked about the need for openness, open interfaces, APIs, which will spur innovation. We also talked about the need for a very specialized silicon to handle the complexity of the math for the physical layer of 5G. And what’s unique about the OCTEON Fusion product line is that it brings those two together very, very successfully and uniquely. It allows our customers, the OEMs who build equipment for the wireless telecom networks to use off-the-shelf standard merchant silicon that has dedicated set of accelerators to deal with the complex math. At the same time, it has open interfaces that allow them to innovate and differentiate. So, two customers of ours can develop very different protocol stacks that have different capabilities because what we do, we provide the underlying accelerators for 5G, and provide open interfaces to use those accelerators in different ways and innovate and build solutions that are competitive in the market.

Ron Westfall: Yes. And I think that differentiation is coming across quite clearly. I think that it’s important for the industry to understand that when you have silicon, you really need it to align with your very specific requirements. And it’s just no longer acceptable to really have to have silicon that is adopted. And then it just doesn’t really conform to what you really need. And I think that’s a very valuable takeaway and know, explaining what is so the distinct and different about the Marvell OCTEON line. So with that, I think it’s also important to understand what else does the Marvell OCTEON line bring to the market. And can you talk about how OCTEON Fusion avoids tradeoffs in features and performance and at the same time support the open scalability of O-RAN and vRAN technologies as well.

Joel Brand: Yeah. Maybe we should address this question in the context of virtual RAN, which is kind of the talk of the town nowadays. In the context of vRAN, the idea is to use commercial off-the-shelf server hardware and use these accelerators to run the physical layer for the 5G, again, because of the complexity. And what we have done again uniquely, we’ve built a dedicated silicon but we’ve used an architecture that mirrors the cloud architecture that is ubiquitously deployed for TCP/IP networking.

So if you look at the cloud environment, they’ve had to deal with a similar issue around IPsec networking, a lot of the traffic is encrypted, protected using IPsec. And there was a need to offload that type of traffic off servers and approach was eventually to use DPU’s and SmartNICs where there are standard interfaces to communicate with these devices, network interface cards, and offload a big portion of the protocol stack processing. What we’ve done… we’ve mirrored that architecture. We are bringing essentially a cloud-native architecture to the radio access network environment by mirroring that by developing instead of a SmartNIC for IPsec, we are developing a RAN-NIC, RAN Network Interface Card that allows you to offload the 5G processing stack such that you don’t need to offload the server. And by doing that, we are avoiding the tradeoffs that are natural in a standard general purpose CPU environment.

Ron Westfall: Yeah, that’s a great insight. I think that’s a very valuable tutorial you provided there, Joel, in understanding the vRAN dimension, and how it definitely interplays with O-RAN architecture builds and adoption, and certainly Marvell’s role in all of this dynamic development and what is going to be important in advancing the overall 5G ecosystem. And with that, I think wraps up the main points of our conversation today. I want to thank Joel for spending time with us, and hopefully, this will wet the appetite of folks out there to find out more about O-RAN technology itself, and certainly the Marvell portfolio and what it uniquely brings to O-RAN innovation, as well as advancing the overall 5G ecosystem. And with that, thank you again all and see you next time.

Joel Brand: Thank you very much.

 

About the Author

Ron is an experienced research expert and analyst, with over 20 years of experience in the digital and IT transformation markets. He is a recognized authority at tracking the evolution of and identifying the key disruptive trends within the service enablement ecosystem, including software and services, infrastructure, 5G/IoT, AI/analytics, security, cloud computing, revenue management, and regulatory issues. Read Full Bio.