The Critical Lowdown Podcast Episode 16
Making the Switch: Why Service Providers and Telcos are Embracing Open Disaggregated Networking
CRAIG SCHAeFER, WILLIAM GRAVES - IP Infusion
Total global mobile data traffic – excluding traffic generated by fixed wireless access – reached around 67 Exabytes per month by the end of 2021. This is projected to quadruple to reach 282 Exabytes per month in 2027. Demand for data shows no signs of slowing and network operators are challenged every day with the need to deliver massive scale and new services—all at lower costs.
One solution, which is proving successful, is a switch to open disaggregated networking. So why are service providers and telcos making the switch?
This is episode 16 of The Critical Lowdown.
Recently our Senior Systems Engineer, and podcast regular, Barry McGinley sat down with IP Infusion, a leading provider of network solutions for telecom and data communications operators. IP Infusion deliver open, standards-based solutions that span from the edge of the network, through the transport layer to multi-cloud environments. With over 20 years experience, they are a proven leader in delivering validated, carrier-grade solutions, with hundreds of customers and thousands of deployments. Let's jump right in.
Subscribe to The Critical Lowdown from EPS Global wherever you get your podcasts:
If you have any questions about or need advice or tech support for your upcoming project, don’t hesitate to get in touch. Or check out our Service Provider Solutions here.
[CS] Thanks for having us my name is Craig Schaefer, I'm with IP Infusion and I lead our Channel Sales organization globally, happy to be here with you today.
[CS] And I'm William Graves, Lead Solution Architect in our channels for IP Infusion and looking forward to having this conversation with EPS.
[BMG] Thanks guys. I'm going to jump straight in and start getting into some questions. We've seen the move of Open Networking from the hyperscale data center to telecoms, which IP Infusions are concentrating a lot on now.
What are telecoms looking for from Open Disaggregated Networking?
[CS] I think it's a few things actually.
They're looking for a solution that can really be custom built for their network. That's one of the things that Open Networking or disaggregation lends itself very well to, from selecting the silicon you want, the hardware you want, to the software you want. Being able to optimize it for the Service Providers use case is exactly what people are trying to do.
Secondly, Telcos and Service Providers want to avoid risk, and that plays very well into what we're doing. We have the ability to deliver these one hand shake solutions that are backed up with our support which is fantastic. You don't have to go source things from different vendors, you can come to IP Infusion and EPS and get pre-integrated supported and an end-to-end solution.
Lastly - interoperability. The solution must be interoperable with the infrastructure that Service Provider has already invested in, and as you introduce disaggregation into the network, how are you doing that because you can't go to the Service Provider and say “hey let's go rip out all of this stuff over here and put my stuff in”, that doesn't work. So being interoperable is that third leg of the stool that we see every day.
[BMG] The one other part I'll add to that is around chip shortages, being able to move to other supported hardware. I don't know exactly how many different hardware vendors you support in IPI, but the ability to actually say “I can’t get a box from Company A for six months”, but I can get a similar box with the same specs from Edgecore or UfiSpace in two months, is a game changer. From EPS Global's point of view as a distributor bordering on integrator, that has been hugely important during this particular chip shortage.
[CS] We've got cases that we can talk about where that supply chain diversity really helped out. The more doors you can knock on to get hardware, the better off you're going to be. That's really played to our success as well.
[BMG] Why is Supply Chain Diversity so important to telecoms or XSPs?
[WG] I think they are looking for some value, they had their existing vendors, a lot of them are getting to the end-of-life areas of the current generation of platforms, and they've been watching the market evolve because obviously merchant silicon is a very big piece of the market, whether it’s legacy or whether it’s open, and I think they're looking for value and options in the network.
The use cases are still the use cases, meaning that there's a lot of interest in the build outs with 4G and 5G, up networks with a lot more capacity; broadband aggregation - in a good chunk of the world it is top of mind because of all of the new access technologies that are being put in a lot of these networks are also being upgraded as well. They may have kicked the tires years ago and open networking wasn't quite ready, but now they're evaluating vendors and they're considering a two-vendor strategy - a strategy that includes a legacy vendor, and an open vendor. It's interesting to see the dynamics, but the bottom line is they're all looking for similar functionality across the board to support their use cases, and they're looking for who can provide the best TCO when it comes to the deployment.
[BMG] We're definitely seeing it on the broadband aggregation side, the money been spent in the UK; Spain are a bit ahead of everybody; but Germany are now coming in for XGS-PON; we're seeing it with cell site routers, which brings me on to my next question.
I was looking at your website recently and there was a blog discussing a few areas that IP Infusion are looking at strongly. The first is Disaggregated Cell Site Routing. There's multiple use cases for it, they call it Disaggregated Cell Site Routing but in fact it doesn't necessarily have to be at the cell site.
What is Disaggregated Cell Site Routing and where is it used?
[WG] We'll call it it Disaggregated xHaul (or crosshaul), internally we describe it as Open xHaul. Particularly with 5G and the introduction of the split between the radios and the Distributed Units (DU's), you now have 3 areas in the network that backhaul or some sort of transport is required for them to support the radio access.
One of those, and probably the most prevalent now is fronthaul, because before that was kind of a high proprietary area that was locked off between the radio vendors, but now with the open radios and the ability to drive everything on eCPRI which is an Ethernet-based form of delivering the traffic to the DU's, I can split my radios and pool my Digital Units that I need to do the baseband processing of those. We see a lot of investment happening in the fronthaul and that's particularly in the areas where 5G is being put in today.
The next area is your midhaul. In these distributed networks depending on how you like to describe it, midhaul might take the areas of DU’s and push them back to the centralized units along with their EPCs (Evolved Packet Core) depending upon where they're located. What you have is a diversity of locations that centralized RAN or OpenRAN starts to drive in those networks.
Last but not least you have good old-fashioned backhaul, I don't want to call it old-fashioned because obviously it's going through a serious bandwidth upgrade as time goes on, but this is the one where we're doing the transport from the Cell Site to the EPCs. What you see in those is that a lot of the processing of the radio is there at the cell site, and that tends to happen in less dense areas. You've got a multitude of transport options that are required, obviously the fronthaul being the most latency sensitive, midhaul being somewhere in the middle, and then backhaul being the existing functionality that people were used to.
We're seeing traction in all three, based on the platforms that we have, and this is the area that we typically have longer evaluation cycles and test cycles because of the majority of the customers here are your fairly large wireless providers around the world.
[BMG] These products have been hugely successful. We saw with the Telecom Infra Project that there are a large number of actual devices now, and the different port counts on the devices are also huge, in terms of Cell Site Routers - Edgecore have 5, and UfiSpace have 7 or 8. From a technical point of view, IPI started off with a data center SKU 5-6 years ago, how much work or R&D has gone into software compatibility on all these devices? You’re about to release OcNOS 6.1, I believe there is a lot of devices supported under this.
[CS] I wouldn't be able to quantify the man hours of engineering time that we have spent on this, but I can tell you that we have been concentrating hard on the Service Provider space, whether it's DCSGs (Disaggregated Cell Site Gateways), routing, but also the broadband aggregation play is huge. The AltNets that are rolling out fiber in the UK with all of this kind of Venture Capital (VC) / Private Equity (PE) investment and it's been a really exciting time.
5G and Edge Computing throws a lot of pressure onto the aggregation points in the network. A lot of the focus and investment has gone there, but at the same time we've still got the products that play well in the data center. As we go forward, you are going to see a pretty good strategy in terms of how we're going to address data center, and continue to address our data center and customers. We've had a lot of success there and at the end of the day we want to go to where the market is going and that’s where the Telco and Service Provider Market is, we're chasing it and that's where we're going to go. But again the data center solutions are still very robust and well proved.
[BMG] We'll go back to broadband aggregation William. We're doing Lots in the UK with this as the pot is enormous, France have a large pot there, Germany are announcing large pots all the time. We have one particular deployment with a company aggregating and the actual OLTs, but there's lots of different things for Broadband aggregation if you want to give us a little bit of an explanation on it.
[CS] I can give you kind of a worldwide view. You’ve talked about the investment that's happening in the UK and Europe, but you've got a tremendous amount of investment happening in the U.S and in Canada. A lot of that tends to be driven by government initiatives, but there's PE floating around in there in terms of the of the companies that are putting this in the ground. You kind of have 3 types of traffic that are being aggregated.
Some of it's active ethernet, which is just standard ethernet all the way to the premise you see a lot of that in the business areas.
The second one is from the cable side of the plants, where a lot of the DOCSIS is being converted into R-PHY or R-MACPHY, which is sort of a fibrant based backhaul for that particular traffic.
And then PON, whether it be G-PON, XGS-PON and maybe the next two versions that are starting to come up as those get conversed. You just basically have a lot of bandwidth being up updated on the edges. I'll give you an example I'm pretty sure the correlate exists in Europe. If you looked at the US 5 years ago and looked at what xDSL coverage that we had in the US and you compare it to today, you see how much that coverage is now fiber. Fiber is driving everything.
That's a huge delta in terms of the amount of bandwidth that I could push back, in some cases 2x, 4x, 6x, to even 10x more in bandwidth that needs to get pushed back to the pot. This is tremendous adjustment for these folks, but there's that investment there is driving all of the application for streaming, over the top delivery of video, etc., and I just don't see an end to it. I think as these infrastructures upgrade or continue to upgrade, and as long as there's money to be made there'll be folks in the in investing and driving this.
[BMG] I have 1G fiber in my house, I don't think I need it, but they will build things for me to use and that's the point. I was actually playing the new NFL game on the Oculus, I can imagine that’s all live and you're using plenty of data with that…
[WG] There's one thing I think that we overlook and I think the pandemic really pointed this out. People looked at broadband as having a cell phone, a convenience device, but I think with the pandemic everybody realized how connected we actually are. If you look at broadband then it could be considered the fourth utility, it's just like having electricity, or water, or heat. These are the things that you need in order to get by in the modern world these days and I think that's also something we should keep in mind in the sense that people aren't doing it just for the sake of driving profit, there's initiatives coming from various governments and so forth to connect the population in a manner that allows for some additional efficiencies that we didn't have, and also to deal with these kind of things that might happen in the future such as a pandemic (hopefully we don't have anymore those), but that we know we can manage to make things work even if we can't move around as much physically.
[BMG] I think the pandemic sped it up, but it was happening, but the pandemic sped it up with the move to work from home. I want to go back to the data center because it is a part that kind of frustrates me with Open Networking because over the last decade we've worked so hard, grand hyperscalers did it, that's brilliant, Amazon and Facebook used bare metal; and Microsoft Azure, but then we did all the evangelizing and we were at OCP events and TIP events and then once it got to a certain point we moved to the Telecom area! It doesn't mean we haven't stopped doing it and we're kind of refocusing ourselves a little bit on it and obviously there's, SONiC, there's still multiple there's still multiple NOS vendors out there, IPI, SONiC, I heard you mentioned data center. Is that a move for you guys to go back and concentrate on that or have you been concentrating on it?
[CS] I tried to address it earlier in terms of the fact that we continue to prove out our robust solutions in the data center. We've got significant large customers and customer opportunities that are on the table as we speak. I think the vibe that we've been giving off over the past couple of years is that we're all about Telco and Service Provider, but what you're seeing more from IP Infusion these days is that our investment has been there over the past couple years, but it's also been in sustaining and making sure that the data center solutions are right there alongside it. That has been the subtle shift in strategy, we're still very much in that game but yes we're moving probably a little bit faster on the on the Telco side.
[WG] If you look at the at the merchant silicon changes, there have been more drastic over on the on the Service Provider side so obviously it's needed to drive a lot more of our attention to take advantage of that wave of new silicon coming out, and if you start to look around the network you start to see some patterns emerge: EVPN and VXLAN, where people are using it both in a data center as well as in Service Provider deployment environments. When you start to see these technologies start to converge there's a natural inkling to where we go and it start to push forward. I think if anything you could characterize our lack of marketing more than anything but not our lack of focus.
We expect to see some upping of that particular aspect in the very near future and we're still driving it with our partners, our partners have been excellent as far as driving these solutions to market. We tend to focus on what we maybe look at a little bit more direct versus what our partners are doing, but our partners are doing an outstanding job in this area to drive revenue and sales in this critical market for us. The whole aspect of it really comes down to sort of speeds and feeds. There's another generation of speed on the way and we'll be involved in that generation. A lot of times what you were seeing in this in the SP market was you had it diverging upwards and downwards, we had firstly came out with the open types of networking it was like in the mid-range, and now we've seen us go downwards toward smaller devices now I'm as low as 32G to larger devices like in the 4.8T terabyte scale type ranges; whereas data center already had that swagger if you will all that range already out of the gate and now what you'll see now is the next generation of data center where we started to bump up the range on that side as we start to move forward. I do believe that we'll be equally focused on both of those as we move forward and that convergence is going to continue to drive that Focus.
[BMG] We’re starting to see some 400G switches now, it'll move to 800G then 1.6T capacity-wise, and it'll just keep going and going. This is because all the good work has been done, the Spine and Leaf is there and it works.
Moving on, there was a section on a blog that I read on the IP Infusion website about Rip and Replace. How are you seeing Service Providers do this? Ripping and replacing may not suit the circumstances, incremental progress might suit better, but which are you seeing more of?
[WG] Depends on the opportunity. We see with a lot of the AltNets that are typically Greenfield, they can just go out and roll it in. But what you see with the traditional providers who have tremendous amount of investment particularly at their edges and in their core, you have to do some interoperation there, there's just no way around it. What you'll find is that maybe they they're rolling out a new 5G Network, it's an overlay from the cell site perspective, but that overlay rolls back into an edge and core layer that's already there. They would deploy the new products in that overlay and then they would integrate them at the edge like they normally would if it was existing.
That's typically what you have there. A lot of times Brownfield means “Oh I have this old box from so and so, it's an end-of-life, I really don't want to change my network but I'd like to see what the new options are to replace it". So we may be going into a network that has a mix of platforms that are aging out and we will be aging those out one by one, and integrating the new platforms in, and that's a sort of onesie twosie replace as you go situation. A migration strategy has to happen alongside. The bottom line in Service Provider is unless it's a brand new network, it's always a Brownfield deployment, there's never really been a much of a full Rip and Replace scenario, at least not in the last 20 years.
[BMG] Which conversations do you prefer?! The new AltNets or the Brownfield ones?
[WG] Whichever conversations are providing a PO!
[CS] There's not that many Greenfield networks out there, those are obviously exciting, but for everyone else there's this opportunity to really test out, improve disaggregation. More than 70% of Service Providers have some type of disaggregated solution in their network, whether it's a more on the PoC or in the lab, but it's not just tire kicking anymore. These things are very real and that's been the biggest thing I've noted in 2022, is compared to the last couple of years where some people view this as a science project, in 2022 it's real. We're experiencing a 1,000% growth in our OcNOS software line as an example, as a proof point, that this is happening. It's pretty exciting.
[BMG] We’re also seeing that in the projects we have lined up for the end of this year and next year, it's reaching a critical mass like it hasn't before. In some cases we’re actually seeing guys not needing to do PoCs anymore. Previously we always had to send over some hardware, it's not 100% necessary anymore. If you can provide a virtual test environment and IPI has virtual software to send over, they want to evaluate the CLI, but you don't necessarily have to send multiple boxes for guys to bash away with in the lab anymore.
[WG] I think what you're starting to see is there's a maturity starting to happen in the industry. Depending on the type of provider that you are and your normal processes for introducing new equipment into your network, that maturity starts to play in. If I'm a smaller player and I'm a quick to market, the situation you've just described is very real. If I'm a larger player and I have to test everything, what it's allowed us to do is to get into those labs, to get into those testing cycles, to become approved vendors at those larger carriers. I would say that a couple of years ago like Craig said, a lot of people wanted to come and play with open disaggregated networking, but now we're starting to hit the mainstream, and are being seriously considered for full-blown deployments in a lot of these large networks.
[BMG] EPS is working a lot with IP Infusion in Africa, the UK, over in the U.S. What deployments has IPI got underway in the US?
[WG] well as far as the ones that are going out of the door very quickly, they're very broadband related. That's been the most active area as far as things that we have gone from ship to deploy. We are in a number of active evaluation and trials with a lot of the larger wireless carriers, and that process just takes its time. It's very similar to the larger carriers you guys have in Europe but I would say a few years ago we weren't even there, now we're there and we're just working our way through these things. We're not ready to brag about anything yet, I think except for our ability to drive the broadband and be part of that revolution in the in the US and Canada, and maybe we'll be able to brag a little bit more next year after we get through a lot of these lab trials and evaluations with the larger carrier!
[CS] From a global point of view, William has mentioned our successes in North America, certainly there's a lot of opportunity in Africa right now, and we've had great success in South America as well. Emerging Markets throughout the world have been ripe for these solutions, in fact in Asia we've had a lot of activity, we've just announced a press release with Lintasarta, one of the leading carriers in Indonesia that we've had success with. It's widespread. That's one of the things that's most exciting. It's not just Tier-1 Telcos, or Tier-3 and Tier-4’s, and it's not just in one geography - it's everywhere! It's a proof in point that disaggregated networking is maturing and ready to take off.
[WG] Barry, you brought up data centers, we've had a lot of uptake of the data center products in South America and we believe that it's just getting warmed up. In the South American market they haven't started a lot of their Broadband expansions, however we're starting to see those ramp up as well. Like Craig says, we're seeing three themes worldwide:
- Data Center;
- Backhaul or xHaul.
We're trying to play actively in all three of those things worldwide, that is what's driving the growth of all of the networks as we move forward.
[BMG] I think that rounds it up quite nicely on those three areas. So thanks very much Craig for your time and thank you very much William.
[CS] Likewise Barry thank you very much.
Glossary of Terms
- DOCSIS: Data Over Cable Service Interface Specification (DOCSIS) is an international telecommunications standard that permits the addition of high-bandwidth data transfer to an existing cable television (CATV) system. It is used by many cable television operators to provide cable Internet access over their existing hybrid fiber-coaxial (HFC) infrastructure.
- DUs: Distributed Units
- eCPRI: evolved Common Public Radio Interface
- EPCs: Evolved Packet Core
- TIP: Telecom Infra Project
- R-PHY: Remote PHY
- R-MACPHY: Remote MACPHY