Deconstructing the Legend of the 5G Surgeon.

In the pantheon of 5G use cases, the legend of the surgeon who performs complex medical procedures from thousands of miles away over 5G connectivity is second only to that of the self driving car in it’s potential to inspire awe. Perhaps I should skip the obvious question where one would be tempted to ask if the presence of skilled surgeons in the flesh is harder to accomplish than all the precision tech they would need, to do what they do remotely, the risks notwithstanding. But I get it, the 5G surgeon is a bit like the Tesla Roadster that’s orbiting the Sun. It can be done of course, but to what end? And how exactly is 5G helping the surgeon’s cause? The answer we’ve been told, lies in lower latency. And this is achieved how? Edge Computing and Network Functions Virtualization, is the buzz.

Consider the journey of a data packet that leaves your smart phone, intended for a destination somewhere out there in the Internet. The packet hops on to the radio waves from the phone’s antenna and rides all the way to the antenna on the tower of your service provider. At some point the packet jumps from the mobile radio network to the IP backhaul network. The basic network architecture topology models on your certification guides or the use cases on the sites of any leading telecom gear vendor out there would tell you that this happens at the Provider Edge router. This PE, in fact puts the “edge” in Edge Computing, literally. The packet then travels to the Packet Core with a couple or more router hops in between, depending on the size of the network. From the packet core, the packet then heads to the vast unknown, the Internet, over router hops again. Over a decade and half back all it took were a couple or two Cisco 7600 routers and a few Juniper MX routers to cater to all the fixed and mobile traffic of a state the size of Kerala and there was no direct internet peering in the whole state for the average provider too. That’s to say, the packet crossed state borders just to reach the promised Internet land back then. This was the case for most Indian states, except for the ones that housed the data centers with Internet peerings. Jump to present day and you have FTTH and 4G in every state and the number of routers I just spoke about sounds primeval, pre-historic and wouldn’t be able to handle the traffic from a single second tier town – in terms of subscriber count and usage trends. Consider this too, one of the fastest growing markets out there, India, is yet to auction the 5G spectrum. I guess we have to make do with 4G and maybe even 3G surgeons over there for a while now. To digress a bit for the sake of perspective, it’s into this very latency sensitive global market that Elon Musk has ushered in Starlink and has managed to stay relevant. But yeah, you could argue that the intent and purpose of Starlink is something else entirely, that is to serve those nooks and corners of the globe beyond the reach of ISPs, in principle at least.

Harking back to the 5G surgeons again, what difference would it make to them at the end of the day if the last mile is over 5G or Wi-Fi or plain old wired connectivity, when they are in fact sitting thousands of miles away? The only time edge computing would actually give the surgeons an edge in terms of latency is if they were, say, in the same block or in the same pin code area or maybe even the same town as their patient waiting under the robotic arm, both hooked to the terminals of the same provider, to be specific. The packets would then hop back and forth between the source and the destination at the edge of the network over 5G without ever having to venture out into the network core or the Internet, fulfilling the edge computing prophecy in its true sense. If the surgeon is out there in the Internet, outside the ISP network, the difference 5G or edge computing would make is zero to none and latency remains more or less the same. As long as these facts go for a toss in the marketing narratives of Telco OEMs, the ISPs or CSPs- as they are called these days- would end up converting their PE locations into full blown Data Centers as hardware in the shape of MEC, NFV and Caches, to name a few, pile up around the Edge Router, and to actually kill the latency, every PE has to have an internet peering too and that’s Valhalla right there for the user, if you ask me. This also calls for a PE device with greater port density. Add to this the mounting real estate, power and cooling requirements at the edge location these additional hardware would demand and that ultimately, inevitably and obviously would reflect on the capital and operating expenditures, not to speak of carbon footprints, energy efficiency and what not.

A decade back, the sheer number of operators in the Indian telecom sector turned it into a dog eat dog situation and mind you, this was mostly over voice services. Now with richer portfolios of content and consumables in the form of streaming services and apps, it’s a different game altogether. The entry of a single giant player armed with cutting edge tech and a product portfolio that tempted the most cynical amongst the customers saw consolidation to an extent and even mergers in the camp of competitors in the very same market. None of these Indian operators would seek the 5G surgeon out anytime soon and would be content with the friendly neighborhood medicine man for now, so to speak. I have been an ISP guy all my life and maybe that’s one reason why I talk the way I do. I might sing a different tune if I switch camps career-wise to any of the gear makers at some point, I admit, in the same breath, I believe that there’s always a middle ground where everybody wins. Don’t get me wrong here now, 5G, MEC and NFV have immense potential and possibilities. Private 5G networks in industrial environments are redefining factory floors and work sites as we speak and Industry 4.0 is the proverbial gold mine for everything 5G.

There’s of course no denying the fact that the greater bandwidths that 5G brings to the table can put a smile on the faces of end users and the ISPs too, but that can be done without emptying pockets on edge computing is all I’m saying here, in so many words. Spending on better coverage and greater network capacities could make those smiles broader too. The perfect user experience in a constantly evolving and rapidly advancing industry, ultimately is a derivative of a shared vision between the technical, business and the financial think tanks of the ISPs. The deployment strategy that might be perfect for a country like Singapore where 80 percent of the network traffic is from indoor mobile traffic (Source- Ericsson 5G SA Case Study for Singtel), would be far from a fit, for a country like India where the ISP would have indoor and outdoor traffic acting as contributors in equal measure – housing trends, geography, matter. These factors further complicate the perspective of network architects. To sum up, when it comes to ISPs and their core customer base, to put a spin on Nolan’s line from The Dark Knight, low latency 5G networks riding on MEC and NFV are what they deserve, but not exactly what they need right now.

The Cloud – Edge Paradox.

It’s quite the paradox to me. You have the whole world, not just the tech world abuzz with the term cloud. Except for the cloud service providers, every business seems to be poised to move into a cloud infrastructure at some point. Unless AWS is going to host Azure or vice versa somewhere down the road. Okay, bad joke. Nothing paradoxical there too, but that’s only until you swipe to see the other buzz about how 5G is going to kill latency and enhance user experience for applications ranging from online gaming to industrial robots to everyone’s favourite ; self-driving vehicles.

Without the hype around low latency, 5G is just 4G with more bandwidth, if you ask me. And how does 5G achieve this “edge” over 4G exactly? That’s where edge computing is ushered in, in all it’s glory. You basically bring your application servers closer to the network edge, literally a hop away from your radio nodes. For perspective, consider the predicament of the friendly neighbourhood online gamer, okay they’re not the friendliest lot, I give you that. Still, for 5G to enhance their FIFA 2021 agile dribbling skils the gameplay servers need to be hooked to their provider edge network and their teammates have to be on the same network preferably on the same provider edge device for a cumulative gameplay experience enhancement. They might as well move in together. This is a broad stroke of an example, for the sake of brevity and though I might be missing the trees for the forest entirely, this is the paradox that presents itself to the network guy that’s me.

Mobile edge computing is basically about distribution of applications servers but you need to distribute your core services as a provider too. It defeats the whole purpose if a user packet has to travel all the way to the core and then head back to the distributed service at the edge of the network. So unless you’re in the business of selling old wine in new bottles and actually intend to walk the talk, you end up investing in a lot of distributed tech for 5G latency apart from the 5G infrastructure. Your edge network port density has to live up to the call for every other cache server out there and you need to collaborate with every game provider, every IoT service provider, every self driving AI provider out there to host their services on your edge network. And your user internet traffic still has to traverse the core network. Meanwhile your cloud is getting lonely up there in your data centers. Maybe you could bring those cloud servers to the edge too. Okay, that’s all far down the rabbit hole I’m gong to go for now.

Now I realise that I’ve said what I said at the risk of sounding myopic and ill-informed but these are some honest apprehensions I’ve had for a while as a network guy. I wouldn’t think twice if I was a product guy, maybe. I would sell cloud today and MEC tomorrow with the same passion to the provider I guess. Yes, you could absolutely argue that cloud and MEC caters to entirely different segments of the same business but the network is where both converge at the end of the day. From a network perspective these are two contradictory design philosophies, so to speak and both are here to stay too. The provider here is forced to adopt both in keeping with the industry narrative driven by technology vendors to stay relevant, ahead of the competition . In the mid 2000s, you almost lost count of the number of telecom operators in India, compare that today’s scenario where the entry of a disruptor in fact resulted in the merger of two leading players and they’re yet to catch up with the game where the provider is not just selling bandwidth and talk time anymore. The jungle has literally turned into an ecosystem and the fittest and the fattest survives. A giant like Jio cushioned by petro cash can afford that, others could just ask the cloud architect, the network architect, the radio planner and the business head to sit down at the table and take a hard long look.