Prakash Sangam:
Hello everyone, welcome back to another episode of Tantra’s Mantra where we go behind and beyond the tech news headlines. I’m your host, Prakash Sangam, founder and principal at Tantra Analyst.
Today, we are back to discussing my favorite subject, that’s Open RAN. A few weeks ago, a Canadian operator TELUS made a big splash when they announced that they will deploy a large scale Brownfield multi-vendor virtualized Open RAN network. I know it’s quite a mouthful, but each word there has its own meaning. We’ll talk about that. And more interestingly, unlike other brownfield operators, TELUS said that not only they will have their new builds with Open RAN, but also they’ll replace their existing legacy network with Open RAN. One reason being Canadian government’s mandate to replace Chinese infra from their network. While that provided a great opportunity, I think TELUS’ all-in move to Open RAN is a bold one and it’s highly commendable. If done right, I think TELUS could pave the way and kind of construct a roadmap for many brownfield operators who are looking to deploy Open RAN in the future.
So TELUS announced that its network will be multi-vendor, as in servers are coming from HPE, chips are coming from Intel, RAN software and radios are coming from Samsung and so on. However, they did not specify their other radio vendor. I think in terms of multi-vendor RAN, integrating RAN software with multiple different radio vendors is the key. So, we’ll talk about that. And when you’re undertaking such a large challenge of deploying a large-scale network with the relatively new technology, of course there are challenges and of course there are equivalent rewards as well. And to discuss all of those with us, we have a special guest today, and that’s none other than the CTO of Telus, Nazim Benhadid.
Nazim, welcome to the show.
Nazim Benhadid:
Hi, Prakash. Thanks for having me.
Prakash Sangam:
We’re really glad to have you on the show. There are a lot of questions. So let’s start with some quick introductions, Nazim. Could you tell us your background? I think you’ve spent more than 20 years of Telus starting at a very entry level and reaching this peak of being a CTO. It would be interesting to hear your journey during those times.
Nazim Benhadid:
Yeah, absolutely. I am a pure TELUS product. And funny enough, that was my intent when I graduated. I really wanted to build a career with one company so that I can see at the end of my career the effect I would have had. TELUS was that. So hired fresh out of school, just to give you a little bit of history on the Canadian telecom market.
It was in early 2000 with the deregulation of the market and Telus being a Western based carrier who was the product of the merger of several regional carriers, was looking at expanding in Montreal and Toronto. At the time, I did my capstone engineering project on MPLS and turned out TELUS deployed the first MPLS network in North America. And since then, I worked in IP, legacy voice, access, mobility, cloud and culminating now with my current role of CTO.
Prakash Sangam:
Perfect! So that gives you an all-rounded view of the business and the technology and set you up for the CTO role, right?
Nazim Benhadid:
Absolutely.
Prakash Sangam:
Perfect. So we talked about your announcement in my monologue. So what is the current status of your Open RAN VLAN deployment? And what has your experience been so far?
Nazim Benhadid:
It has started and we’re ramping up. So far, the experience has been extremely positive. So to recap, every one of the component is from a different vendor. So far, the hardware is from HPE with an Intel chipset. The container platform is from Wind River. The network function is from Samsung. And the ORUs are both from Samsung and from a third-party Canadian antenna provider that should not be named yet because we’re still going through our procurement activities. But we’ve tested both with Samsung radios and non-Samsung radios. We’ve deployed in production, and so far the results are as good or better as traditional RAM. So we’re very, very pleased with where we are.
Prakash Sangam:
Perfect. We’ll go into a little bit of a detail as we go through the interview. So to start with, what was your biggest motivation for choosing, you know, virtualized Open RAN?
Nazim Benhadid:
So there’s a couple of dimensions to this. There’s a strategic dimension and there’s a tactical opportunistic dimension to it. So on the tactical opportunistic dimension you already mentioned, we have to rip and replace our network. And if we are to do that, we thought that it would be wise to replace with the future technology versus like for like.
So, you know, we have to do this replacement on our own dime. And there’s really very little value for us by doing that. So by going into RAN and getting the benefits of RAN, we think we can make good of an unfortunate situation. But there’s a more strategic conversation here, which is the competitivity in the OEM ecosystem.
We were already a consolidated industry, but with what seems to be more and more a split of, I don’t know if I should call it French shoring or having some vendors allowed in some parts of the world and some vendors not allowed in some parts of the world, create a situation of not monopoly, but it reduces competition.
Open RAN actually not only tackles this issue, but it creates opportunities. So now, you’re an expert in this domain, you know that when you go into a traditional RAN vendor, you’re kind of locked in. The substitution costs are very, very high. Once you make your decision, you’re kind of set for it for a period of time, regardless of the performance or the innovation of the vendor you selected. With Open RAN, you have a lot more optionality.
Prakash Sangam:
And in terms of cost, was that a consideration for you guys? And also, when you’re starting to deploy and you place orders and so on, are you looking at the cost being similar to a legacy, higher or lower than that?
Nazim Benhadid:
So it’s lower than that. And we believe that we’re at the beginning of the curve and that future costs are going to be lower. Not only costs of deployment, but also operational costs. Both the care and feed of the network, RAN comes with a strong framework for automation. But also in terms of innovation, if you look at one of the big costs of a wireless network is power consumption.
With RAN, with the right innovation, it allows you to really, really control how much of the power you use and what does sleep mode mean and how can you optimize that. So we’re bullish on not only the benefits, the TCO benefits that we’re seeing today, but also the potential for further TCO benefits.
Prakash Sangam:
And who is doing the system integration, which is the most challenging part in my view for Open RAN? Are you doing it yourself or your vendor doing it or how is that?
Nazim Benhadid:
So we’re playing a part in system integration. A couple of reasons for that. First of all, if I step back a little bit and look at our journey on the telco cloud, we chose very early to build a multi-tenant open telco cloud, which means that we had to do some serious system integration at the time and kind of developed the muscle for that.
So that’s one of the reasons we decided to do it ourselves. Yeah. But the second reason is also because very early on, we came to the conclusion that we had to pick an ecosystem of partners. And that with the right ecosystem of partners who are open, who are willing to build affinity at the interface between their own systems, well, the integration effort would be much lower than anticipated by the industry.
Prakash Sangam:
And that’s what you’re seeing when you’re actually deploying. Don’t you have to do a lot of changes, other things when you’re connecting these different pieces from different vendors?
Nazim Benhadid:
No, it’s fairly stable. And even, you know, integrating or testing and integrating new ORUs, we’ve fine-tuned the process to be able to do that within eight weeks.So far, we were really nervous about this, but it turned out, if you make the right decisions upfront and you have the right collaboration from your partners, then the integration is something that is, I wouldn’t say trivial, but definitely manageable.
Prakash Sangam:
And you said you did the testing in terms of performance and then you were happy with the Open RAN one. Did you specifically test 64 TR because that has seen challenges initially in terms of providing performance across vendors? That too cross-vendor 64 TR configuration and comparing that to a legacy system?
Nazim Benhadid:
100%. So we’ve conducted first extensive lab testing and then field testing with some sites with over 40 cells per site and we’re satisfied with the results.
Prakash Sangam:
Okay, so it’s like a commercial grid over the year testing, not just lab with one or two sites and so on.
Nazim Benhadid:
No, no, no, we’ve had clusters of over 30 sites in production for a few months now. We look in the wireless world, at least in Canada, customer experience is extremely important. It’s a competitive differentiator, and we couldn’t afford to go into production with something that wasn’t extensively tested in production. And so far, as I said, we’re very, very happy.
I mean, we did run into some things that needed adjustment. So for example, you know, external temperature in winter. But our experience is our ecosystem of vendors between the open radio and the rest was able to actually build code, deploy in production and fix the issues in less than 24 hours.
Prakash Sangam:
OK, perfect. That’s good. Do you as a learning, do you suggest small, medium size operators that they do the system integration themselves or you would rather advise them to either get a vendor or a third party to do it for them?
Nazim Benhadid:
I think the most important decision for them is which ecosystem do they attach to. And then, you know, if you pick the same ecosystem that that we picked and other carriers, then that ecosystem, you benefit from all the work that we’ve already done.
Yeah, so we basically get the same chip software, hardware, you know, in the cloud. Yeah, it’s already kind of you implemented already.
Prakash Sangam:
So it’s basically repeating the same configuration you’re saying.
Nazim Benhadid:
100 percent. You know, Prakash, one of the challenges of the telco world is scale, especially in a virtualized world. So if you look at, you know, telecom is a fraction of what IT represents in terms of compute and container platforms. Of course. And it’s very, very fragmented. A lot of proprietary, you know, protocols, a lot of the network functions not really cloud native and, you know, still attached to the hardware.
And all of that creates huge inefficiencies in addition to the size of the operator. As an industry, we can only be successful, especially in a world of, you know, plateauing or declining revenue on the consumer side. If we can attach or create scale and attaching to existing ecosystem is really the best way to create scale.
Prakash Sangam:
So you said you are from day one itself, you’re going with the two radio vendors, not just one. So then how are you deciding between the vendors? Is it like a geo based decision that this market will be vendor one, that market will be vendor two? Or is it based on the configuration? How is that mix and match strategy working out?
Nazim Benhadid:
So we plan to have a mix on every site. Depending on the bands and the coverage objectives, we select the best tool for that objective.
Prakash Sangam:
So I mean, you’re looking at performance for that specific band from that specific radio vendor?
Nazim Benhadid:
Yes, performance cost.
Prakash Sangam:
I see. But that’s why I call it True Open RAN, right?
Nazim Bnehadid:
So every one of our sites will have two radio vendors, both of them.
Prakash Sangam:
Wouldn’t that also bring in complexities in terms of managing and so on?
Nazim Benhadid:
It does to a certain extent, but the benefits compensate for that. So when we look at it, we want to be able to get the best TCO per site. And that’s why we’re looking at it this way.
Prakash Sangam:
And you said specifically the decision point between these two vendors is, what is the performance for that specific vendor for that specific band, you’re saying, or is it?
Nazim Benhadid:
Yeah. So vendor performance per band and cost.
Prakash Sangam:
Yeah, OK, because some other operators I’ve talked to, it’s basically they divided between, say, 16 by 16 TR and below is one and above is another vendor and so on, and some have done geographically as well. Good.
So I think the answer to this is yes, but I still have to ask, Open RAN, so all of the interfaces that you have are O-RAN Alliance compliant, for example, if you want to introduce a third vendor in the radios or if you want to bring different RICs or X apps and R apps, you just plug and play.
I’m sure there will be some integration in it, but that is minimal. It’s all fully O-RAN Alliance compliant?
Nazim Benhadid:
It is, yes, 100 percent, although and the specifications, at least on the radio side, are mature enough for seamless integration, but there’s still more standardization that’s required on the RIC side and for X apps and R apps integration.
Prakash Sangam:
Yeah, we’ll come to that in a little while. So I have some specific questions regarding that.
And you are also, because of the situation, it’s a rip and replace of your legacy, which means the 4G network will also be VLAN and Open RAN. And how do you plan to do that without affecting the customer experience?
Nazim Benhadid:
So we have a swap process on a site-by-site basis that is radio by radio. It’s actually quite cumbersome, but I know some carriers will just shut down a site and create a coverage hole for the time that they do the work, but we don’t. So the way we do it is we’ll have the two systems present and then we’ll change from one to another over a number of days.
Prakash Sangam:
So it’s like some bands are still transmitting, you swap the other bands and then turn those on, because otherwise you’ll need twice the spectrum, right?
Nazim Benhadid:
Yes.
Prakash Sangam:
And in terms of baseband silicon choice, as you said, you went with Intel, which has a built-in look-aside accelerator. I think you went with 4th generation, which has this V-boosting as well, 4th generation Xeon. And I know that Samsung also has another option, which is with Marvell, with an external in-line accelerator. Did you look at both? Did the test and decide, you know, what was your decision process like for specifically baseband and Layer 1?
Nazim Benhadid:
Yeah, so we did evaluate various options of look-aside, the in-line accelerator, GPU-based design. But at the end of the day, we came to the conclusion, again, that it’s a matter of scale. So our aim is really to standardize a compute architecture that would work across different network functions.
We didn’t want to have customized hardware. And we tried to go with a CPU architecture that is pragmatic. We had concerns about performance, but as I said, both in our lab testing and our field production trials, those concerns of performance did not turn out to be actual problems. So we’re very happy with our choice.
Prakash Sangam:
And in terms of the system integration, was it mostly interaction and working with your direct vendors, or did you have to go back to Intel as well, work with them to make sure everything works fine?
Nazim Benhadid:
For the most part, we worked with the vendors. I mean, we have conversations with Intel for future-looking projects, but for the most part, Intel supported directly our partners.
Prakash Sangam:
That’s good to know. Okay. So about the SMO part, xApps and other things. So you were the first one to deploy Samsung’s SMO, and how was the experience? What are your objectives? What are they able to achieve and planning to achieve? What are all things you’re planning to use the SMO for? Is it ran only or it’s across the stack and so on?
Nazim Benhadid:
So first of all, as I was saying, first of all, it’s a tool of tools, right? That’s how we think about it. It’s a framework. And some components of SMO as a framework are things that we are already using in our telco cloud. So we really favor horizontal scaling and reusing as much as possible.
And some components are still, they’re still more standardization. So between the DU and the non-real time RIC, the O1 interface is somehow mature. There’s still a gap in E2 between the DU and the near real time RIC. So for now, we’re using what’s available to get us started, but we are not at our final destination.
Prakash Sangam:
It’s a work in progress, in other words.
Nazim Benhadid:
Yeah, a work in progress. In a sense that we’re contributing to the RAN Alliance, the respective groups that have to do with RIC and SMO and all of that. We are also looking at integrating, as I said, existing toolset in a horizontal scaling perspective. And we believe that we still have at least a year before we get to the level of maturity. We can have strong opinions on the best way to do it.
Prakash Sangam:
And are you using RIC and some apps now or…?
Nazim Benhadid:
Nothing worth mentioning for production.
Prakash Sangam:
So like the network going in, it’s basically using the traditional system, not the SMO, the new SMO? You’re using obviously your existing components that you’re using with your telco cloud, but for the new Open RAN system as such, it’s still using the legacy approach, correct?
Nazim Bnehadid:
We are leveraging some of the components of the SMO from Samsung for automation and so on and so forth. But it’s not yet something that I would come to you and say it’s groundbreaking.
Prakash Sangam:
I mean, it’s very early stages, right?
Here is a huge value of Open RAN. Small spotters can come in and come up with interesting use cases, interesting tool chain and so on to improve the efficiency and make RAN network a little bit different than legacy. Otherwise, it’s the same components with different renders, different interfaces.
Nazim Benhadid:
100%. So as much as I like the fact that now the deploy and a big part of the day to support is fully automated, I still think that the value will be in added values in X app and R apps. But the market is still small. You have more and more carriers such as ours that are making announcements around being RAN supporters and deploying it. I think that’s going to create the right momentum in the market to actually start fostering innovation and seeing interesting things in this space.
Prakash Sangam:
Cool. So looking into the deployment and the future, so you said it’s already started. What are some of the milestones in terms of timeline when you’ll have substantial completion and full launch and so on?
Nazim Benhadid:
Yeah. So as I said, we already have it in production. In a couple of months, we’ll be deploying exclusively RAN in our swap. And what’s left of our network is basically 50%. So by the end of next year, we’ll have 50% of our network that is completely all-RAN.
Prakash Sangam:
And you’re saying end of 2025. Perfect.
Nazim Benhadid:
And then hopefully, the other RIC xApps, rApps will grow and then have some sort of maturity next year. So probably more to talk about then.
Prakash Sangam
Okay, perfect. So yeah, thank you very much. Did we miss anything? Any other things you want to point out regarding your experience with Open RAN?
Nazim Benhadid:
Nothing in particular other than, we know what traditional RAN can provide and we know that it’s been very slow, that in terms of innovation, that has high lock-in, has high cost, and it doesn’t scale. So I really encourage our peers in the industry to look seriously at RAN, true RAN.
It’s possible, it works well, and the more that we invest in it, the more scale it will have, and the more it will work for all of us.
Prakash Sangam:
Excellent, Nazim. Thank you very much.
This was a great discussion, and thank you for patiently answering my questions, every little of them, and sharing your insights. And best of luck with your deployment.
Nazim Benhadid:
Thank you, Prakash.
Prakash Sangam:
So folks, that’s all we have for now. Hope you found this discussion informative and useful. If so, please hit like and subscribe to the podcast on whatever platform you’re listening this on. I’ll be back soon with another episode, putting light on another interesting tech subject.
Bye bye for now.
