Ahead of Mobile World Congress, 6GWorld sat down with Susan James, Senior Director of Telecoms Strategy at Red Hat, to talk about the early indicators they are seeing in edge networking and how that relates to other trends such as private networking.
A Different Form of Latency
James started off describing a use case that Red Hat has recently been involved with, which underscores that – while edge computing may be helpful for low-latency applications – it is far from the only scenario where companies of different kinds will find value from being able to access a platform combining elements from the enterprise, cloud and edge computing.
“There are a lot of underserved use cases today,” James commented. She shared the real-life example of a fruit farmer Red Hat worked with to improve his efficiency in taking and using crop data.
“What he used to do is drive past his crop, scanning the field; then he would have to take the data in his car and drive twenty minutes home to upload it to the cloud so he could do his analysis. Then drive twenty minutes back in order to spread more fertiliser or whatever it is he needs to do.
“Essentially they’ve implemented a private 5G site at the location. He now drives through the crop doing the scanning to see what the foliage looks like, how many buds are on the trees, the water in the soil, all that stuff; and basically he knows instantly when is the best time to prune so that the apples are getting enough sunlight and so on.”
The impact can be significant, not just in terms of time saved in transporting data but in being able to provide targeted responses in real time.
“In this example the farmer spent something like five million dollars’ worth of chemicals on crops in a growing period,” James explained, before highlighting the reductions in the use of fertilisers and pesticides that the new approach delivered. “Being able to reduce that bill by half is a massive difference. That’s 2.5 million dollars that doesn’t leave the farmer’s pocket, especially because the cost is up-front before you know what the yield is going to be. The thing I’m even happier about is that there’s 2.5 million dollars’ worth of chemicals not going into the soil, which is a huge environmental benefit.”
Enterprise Data and Enterprise Information
This all sounds like a positive story for private networks, but one of the key elements is how the data is absorbed, moved, analysed, and shared. In this example, the 5G base station also included a Dell SmartEdge server. “It comes with the software, it comes with the SmartEdge application already on top of it and you can just sign a contract with one person and you’re off,” James said.
The advantage here is mainly for the telco, James noted, “because actually moving the data around is quite expensive. As soon as you move data into a public cloud you have a whole host of additional problems such as data security, GDPR issues and more. If you can have that ‘cloud’ at the edge then you have fewer of those issues to solve and you don’t need to ship data around. You just need to move information around.”
The difference between data and information largely depends on context, James observed, but usually boils down to “the important bits that you care about. That you may want to send back to the cloud and store for future reference – the changes you saw [between measurements] and those sorts of things.”
Note that, in this scenario, the placement of analytics workloads in the cloud or at the edge is driven by the operator, not the end user. This has a number of implications for the operator, allowing them to better manage traffic in a situation where data usage through telecoms networks keeps growing drastically. Placing workloads at the edge can delay or remove the need for deploying additional infrastructure to transport data between cloud and user. That’s not the only benefit, though.
“Just to take that further, becoming carbon-neutral and sustainable will require us to take better decisions much earlier, so that we spend fewer resources, we consume less electricity, or we have the ability to optimise what we have,” James pointed out. “That, I think, is what becomes really interesting because it makes good business sense economically and also it makes good environmental sense.”
This, in part, depends upon a consistent platform working across the enterprise, edge and cloud infrastructure.
“It takes a number of companies to do that,” James said. “It’s not just the telco, although they’ll be part of that solution. It’ll be the telco, it’ll be the systems integrator, it’ll be an agricultural firm that specialises in taking these lidar readings and turning them into information.”
Red Hat is, unsurprisingly, also involved, partly by including their OpenShift software, but also more directly in the move towards an information-centred society.
“We just launched Red Hat OpenShift Data Science, where instead of having to build your data lake and so on we do that all for you, so you have the ability to consume that as a managed service. Well, if you want to have it deployed you can, but we actually started as a managed service because getting analytics up and running can take a bit of time.”
For the enterprise as well as the operator, moving from data-based towards information-based concepts can help with simplification and storage benefits. “You probably don’t want to keep the data lying around, you just want to keep the information.”
The Business Case(s) at the Edge
The acceptance of cloud computing, especially during the last 18 months, means that the market is primed for solutions that are not based solely or mainly on the companies’ hardware. James observed, “If you’re an enterprise, do you really see the benefit in maintaining your own hardware infrastructure? I think we’ll see a big shift towards people saying ‘Yes I want this, but my focus is on growing my apples, not maintaining a server. Yes, I need to have it, but is that really my skill set or do I expect someone else to manage that for me? Do I really want to spend thousands of dollars up front, or is it an incremental amount that I pay?’ All these things are being tested out in these early use cases, and lots of service providers are doing that.”
The growing acceptance of working from home brought about by the COVID-19 pandemic also has another side effect. For many businesses, functions such as data warehousing and analytics would be out of reach if it weren’t for cloud-based managed services that minimise the need for local specialist knowledge.
Combining network and analytics capabilities together has the scope to assist operators with some network densification activities.
“When is it sensible to deploy a site?” James asked. The business case for doing so can look very different depending on who is assessing it. “From the telco perspective an enterprise may be the only high-value client in the area. Maybe it’s not the best thing to build coverage out over that area.” For an enterprise, however, paying for the coverage of a private network – potentially as part of a package of services – offers a business case that can be compelling.
That said, fragmentation of business models, technology and the role of the operator will vary significantly in the short to medium term. “The edge will be very fragmented to start with, and then it will consolidate over time. So yes, there’ll be Wi-Fi 6; yes, there’ll be 5G; there will still be a lot of narrowband IoT running on 3G for some time; so I don’t expect that in the next couple of years things will settle out.”
Throughout this process, pragmatism will be critical. Referring to a recent conversation with a telco, James observed that the relationships between operators and intermediaries has always been there, though maybe not quite so visible. However, these will become more and more important in these edge scenarios.
“Quite frankly, [the telcos] don’t care if you know whether it’s them at the back end or whether they are fronting the engagement. Either way they’re in the mix and that’s what they care about. I think that’s how it’s going to pan out more.”
Apples and Edge Computing: A Taste of Things to Come?
Going back to the farming use case – what people refer to as the edge varies hugely. Can or should we expect edge servers to be regularly co-located with base stations, to really minimise the traffic and processing work further back in the network? Quite simply, it depends.
“One thing is the density of the population that you’re trying to hit with the service. In some cases, you’ll actually deploy on a very, very small box far out and in that particular case it’s highly unlikely you’d be deploying anything else on that. But it also depends on how big the application is and are you making it available for multiple enterprises or that specific enterprise?” James explained. “It will change over time, of course. For example, with Radio Intelligent Controllers, X-Apps and these sorts of thing.”
The main limitation is with the servers themselves, James noted.
“In a radio unit scenario, the applications I expect would be very small right now, because the power consumption, footprint, and considerations like that are really important so far out at the edge. Then the question is, does it make sense to put the application there, or is it better on the device?
“Things are going to become more blurry about whether something is the network or not. Even from an operator perspective that boundary hasn’t been as clear as we like to think it is, because they’ve always supported customer devices.”
Working far out at the edge is a challenge, too, for Red Hat. “The further out you get in the network the smaller the server footprint and that means different requirements. We’re really scaling down the size of the deployments.
“Then [there’s] the whole management of that, because it’s not going to be just one server. In many cases these networks are going to become extraordinarily large. We’re talking about thousands and thousands of servers, distributed all over the place. How do you manage that and make sure they stay secure? Those are the sorts of things we’re working with.”
All of this shows movement in promising directions, making sure that the technology, the collaborations across hardware, software and systems integrators, and even the business models for the customers and value chain are all coming together. While this is exciting progress, James offers a note of caution about the speed of adoption.
“Once the technology’s available, then people can start thinking about how to use that technology to solve their particular problems. So there is a lag between when something is possible; when something is possible and it’s in your region; when something is possible, it’s in your region and you’ve understood what you can actually do with it.”
