I think probably the biggest confusion with Open RAN is that there are not one, but two axes that people are looking at.
Disaggregated hardware and software
On one axis is the concept of disaggregated hardware and software. The idea behind that is that you can run virtualized software for performance and run it on standardized hardware. And that’s often why when we write about Open RAN, the term tends to come from that, and it came out of the telecom infrastructure project initiative that Facebook started.
Industry-Directed O-RAN Alliance
The second one is more industry-directed and comes out of the O-RAN Alliance. And what the O-RAN Alliance wants to do is create standardized interfaces between the centralized unit (CU), which sits at the edge of the network, the distributed unit (DU), which performs the layer one processing, and the radio unit (RU).
The idea behind this is that you could use company A’s radio unit with company B’s DU and company C’s CU. And those are the, sort of, separate layers. This is often called OpenRAN and is driven by the O-RAN Alliance, an operator-led standards body and initiative.
Proprietary RAN vs Proprietary hardware
Now you get four quadrants out of this, and you can do all combinations. So you can have a proprietary RAN implementation that runs on disaggregated hardware and software. And you can have an OpenRAN implementation that runs on proprietary hardware. You can do either those two or have a completely open implementation with standardized interfaces but not running on commercially available hardware.
Cloud-Native Architecture
For us at Supermicro in particular, we are most interested in the disaggregated hardware and software because we want to go after that standardized server model. And it’s the standardized server model that ties back to the 5G cloud-native architecture. So, when you put those two together, you get the benefits of 5G Open RAN.
It’s happening very quickly now. Initially, there was a lot of skepticism and different degrees of buy-in as to whether people would really adopt Open RAN and whether it provided the benefits.
Adoption – Now we see that there is an industry-wide adoption at different levels. Ericsson, who has the most proprietary equipment, was probably the last one to come in. But they’ve been fairly open now about that they are going to do Open RAN. Japan is emerging as a leader in actual Open RAN deployments and development, NTT Docomo and Rakuten are prime examples. At other operators, we see more CRAN deployments following the Open RAN model of disaggregated vRAN software running on COTS hardware.
Optimum for specific deployments – Open RAN won’t be the optimum solution in every single case – it’s not that it’s going to be 100% that the model is going to move to Open RAN. And I think once people have started to recognize that, the adoption gets a little easier because it becomes about: when is this the optimal solution and when is the best place to use it.
Open RAN adoption for 4G Networks – Now we’ve seen two pioneers in the Open RAN space. The first one happened with 4G networks, particularly in rural areas. In the older networks, a lot of the voice still runs over the TDM 2G network. So you end up having to run three networks: a 2G, a 3G, and a 4G network, and each one has its proprietary equipment that gets very expensive, mainly where your population density is not that high.
Rural Market – We saw Open RAN being adopted by the 4G industry in rural markets, where the processing power, you could argue, wasn’t too complicated. But it validated the systemic principle and the whole idea, the concept.
Open RAN in Greenfield & Rural Markets for 5G Networks –
Translating that into 5G networks, we’ve again seen the early adoption happening in these more rural markets, but also by two greenfield operators, notably Rakuten and DISH Network. Now for them, it’s easier to adopt new technology because they have no legacy, and they don’t have to worry about backward compatibility or interoperability between the two. So unsurprisingly, they’ve gone ahead of the fastest.
Going Mainstream – Now we’re starting to see it go mainstream, and you’ll hear more and more announcements daily.
We have been involved very early, right from the beginning. At the inception of running the software on disaggregated hardware. Intel was really, the creator and pioneer of this. We have worked very closely with them from the beginning.
Supermicro is also a member of both TIP and the O-RAN Alliance, as one of the workgroup editors.
Many of the early reference designs have been built on Supermicro hardware. And we continue to expand and engage with all the main industry players, the independent software vendors, and several operator networks. So for us, 5G RAN is another edge network application, and it fits beautifully with our standardized server technology. In each successive generation of its Xeon processors, Intel has enhanced the set to improve 5G baseband, networking, and AI processing. The latest fourth-generation Xeon server processors gain new instructions that enable Massive-MIMO capability and accelerate vRAN workload execution.
There has been a holdup in the lack of widely available GPS-enabled timing-sensitive Fronthaul NICs (FH-NIC) that would allow the separate and expensive Fronthaul Gateway appliance to be integrated into the DU. These FH-NICs are now becoming available from several suppliers, some even with integrated accelerators. The combination of these FH-NICs and the latest processors is at the maturity needed for large-scale deployment.
Supermicro will have a broad range of servers based on next-generation Intel Xeon Scalable processors for both 5G Core and RAN.
Subscribe for industry insights. Elevate your influence – promote with us!
