With brownfield operators poised to scale Open RAN, Intel highlights the TCO benefits of Ethernet transport—availability, familiarity, flexibility
A number of converging technological trends are reshaping how operators design, deploy and manage mobile networks. The push to cloud-native 5G Standalone will be key to unlocking new revenue opportunities and automation-enabled operational efficiencies. The distribution of compute out of data centers and to various edges is essential for delivering real-time use cases, including artificial intelligence applications. And the disaggregation of hardware from software in the radio access network (RAN), supported by standardized Open RAN interfaces, is gaining buy-in at a global scale.
While these various network domains all come with sets of unique challenges and opportunities, there’s a quite literal connection between them. Transport perhaps doesn’t draw the same attention as flashy RAN or compute implementations, but there’d be no wireless without wires as the adage goes. In a conversation with RCR Wireless News, Kartik Srinivasan, general manager of Intel Ethernet Products and Solutions, made the dual points that Ethernet plays an increasingly key role in connecting distributed radio access networks and because it is inherently an open, standardized technology with an established ecosystem around it, it’s directly aligned with the concepts of Open RAN as a paradigm shift with lines through to customization and economics as enabled by a flourishing ecosystem.
“Open RAN, by definition, requires any part of the technology that touches its deployment model and strategy to, [again] by definition, be open. And Ethernet is one of those connectivity technologies that has been a standard for over 40 years…The ubiquity of Ethernet has shown as a big value add because availability of technology, familiarity of technology, is extremely important, and is a critical component intrinsic to openness and Open RAN as an extension.”
So in the context of Open RAN as a vector for more programmable networks and better economics for operators, why does that matter? And, perhaps more importantly for the long-term success of Open RAN, how does it map to TCO?
First, Srinivasan explained, because Ethernet has been standardized and available for so long, it has already achieved, through the collective work of many firms, more advantageous economics than proprietary or otherwise nascent technologies. This also leads directly to the availability piece. “And the final point that I’ll make is the flexibility of the technology.” He gave the example of an operator using commercial off the shelf servers to host RAN and other workloads in a distributed architecture, and noted that operators may be leveraging these new architectures to address new types of use cases beyond macro coverage for a public network.
“If you want to tune Ethernet for multiple densities, if you want to tune this for different speeds, if you want to tune this for different connectivity types, Ethernet has a plethora of options that allow the customers–whether it’s server OEMs or the ODMs or the end customers, the operators or the infrastructure providers–it allows each one of them to pick and choose which flavor and which kind of deployment model they would like receive, and deploy it with Ethernet.”
At a high-level, an Open RAN system would split up the functionality of a radio and a baseband into a centralized unit (CU), distributed unit (DU) and radio unit. The CU and DU are essentially commodity servers running virtualized network functions in software; where which functions are run is, to some degree, up to an operator. But, as you’d expect, the CU, DU and RU all need to be connected. Also of note is that Open RAN is intended to be a multi-vendor exercise wherein different parties can provide compatible hardware and software allowing operators to tailor the RAN to whatever their (or their customer’s) needs are. All of the above dictates that any Open RAN deployment requires numerous vendor partners, test specialists, system integrators and potentially more parties, to all work with an operator and with one another.
To give a relevant example, earlier this year Intel worked with Cisco to integrate equipment into an open fronthaul link between a far edge data center hosting the DU to an RU at a cell site. Intel provided its Intel Ethernet 800 Series Network Adapters with enhanced timing capabilities and Cisco brought its Nexus 93180UYC-FX3 network switches. This shows the deployment-readiness of multi-vendor Open RAN fronthaul transport.
Ecosystem collaboration, Kartik said, “is such a critical component for success for Open RAN. A single vendor does not make an ecosystem, so you’ve got to have multiple vendors participate…[and] collaborate…This ecosystem requires a lot of collaboration. And collaboration, not just in terms of interoperability, but collaboration in terms of joint development.”
Specific to Intel’s value proposition in Ethernet and in general, “The fact that Intel has been around for the last 40-plus years naturally lends itself to the available technology components already being, for the most part, interoperable. Not just the ecosystem partners, but compliant with standards.” Back to Open RAN, this means Intel has worked with ecosystem partners, including OEMs, ODMs, ISVs, test vendors, system integrators, module makers and transport vendors on stringent compliance and interoperability testing “to enable our customers to have flexibility in their deployment and availability when it comes to RAN…I’m proud to be part of Intel because we live that culture. That is part of our DNA.”
For more perspective from Intel on Open RAN, check out the following:
- Virtual event: Open RAN Global Forum sessions Ecosystem maturity and evolution and A deep dive into AI in the RAN, available on demand.
