Why the network matters

Since the pandemic there has been an increased focus on digital transformation across the spectrum. Yet many enterprises do not think enough about the network. Jefferson Wang explains the critical value of a modern network and how it delivers a competitive advantage in Industry 4.0.

Below is a transcript of this podcast. Some parts have been edited for clarity.

Michael Hainsworth: An IT (Information Technology) network is different from an Operational Network. As the enterprise accelerates its move to Industry 4.0, it is finding its existing network is not capable of handling the ultra-reliable, low latency needs of next generation sensors, cloud, and AI (Artificial Intelligence). That is where Jefferson Wang comes in. The Global Cloud First Networks and 5G Lead at Accenture is shepherding companies into the future. He tells me the existing network is typically the weakest link in the chain pulling an enterprise into the fourth industrial revolution.

Jefferson Wang: I think a lot of times we are starting these solutions and these solution discussions with trying to figure out what the business problem is and then working backwards. And when you go through that, you do a great job figuring out what the actual business impact is. What is the actual problem you are solving? What type of application will you need? Trying to think about computing, cloud and storage. And then eventually you almost assume that the network is there and ready, but a lot of times in that approach, you do not really dig into the actual network characteristics around latency. Is this more of a video-based solution where you need more down link? Is this more of a situation where you need reliability, and you need to keep the jitter down?

A lot of times that becomes an afterthought, and you will get into these pilots, or these programs where you will start to deploy, and suddenly the network becomes the actual issue or the limiter. Versus designing that in from the beginning, where we believed that the network could be an advantage. The network can be the actual competitive advantage. You design it from the beginning.

MH: You need to know what you need the network for. And then you need to build that around it. A system designed to ensure worker's safety on the assembly line is going to have very different needs than a system that monitors the elevators.

JW: That's exactly right, Michael. And also, if you think about it, there are certain things that different types of networks can do those others cannot. And that's the actual competitive differentiator. I just talked to a client recently who said a lot of my manufacturing right now is in Asia and my competitors are in Latin America. And I talked to them about, well, your operations in Asia now based on location, having access to 5G, having access to spectrum and access to these new networks. And there is a set of OT (Operational Technology) use cases that you can do as you build your new plan, that your competitor will not have access to for another year or two. And that immediately lit this light bulb up with wow, the client said, “so this network that I have, based on just sheer luck of where my manufacturing location is, is a competitive advantage?”

MH: The heart of the matter is, it seems, we are trying to leverage existing IT networks to build out operational technology, OT networks. How would you describe the distinction between the needs and objectives of the two?

JW: It is a great question to really dig into, right? A lot of times we just think all networks are created equal. They are always going to be there, and we can use them for all the same purposes. There is a big difference between an IT network that is originally designed for more data centric computing or just moving data around, but think about your HR systems that need to move data around versus an OT network for operations, which is really focused on monitoring critical elements, measuring important processes, dealing with devices. So that OT network is really mission critical. Your business is dependent upon it. And there is a different level of reliability, security, and availability.

MH: Your example with the client in Asia, is fascinating to me because, with 5G and private networks, one of the key benefits is ultra-low latency, high volume of connected devices, yet the patchwork of network technologies that a lot of companies are already using and trying to deploy, degrades performance.

JW: Yes, it is a big problem, right? A lot of time when a network is an afterthought, the situation becomes, well, let us just use whatever we have and try and fit the solution, the application, or the compute layer into it and figure it out. And the situation you run into is that this patchwork of networks becomes that limiting factor, you have tons of operational expense on keeping all these different network technologies running. You have different upgrade patterns where a lot of times these types of equipment fail at different times. And then you just have overall performance inconsistencies. So, it is really hard to develop solutions that work for this. For example, if you have multiple solutions based on location, take mining as an example, you have an above ground network and you have a below ground network and those are two very different situations. And a lot of times above ground, you could get away with potentially just Wi-Fi, but below ground, you have to think about a whole different technology pattern.

And as a result, those decisions really require something more common and uniform. If you design an application for that, you are going to go to the lowest common denominator, which may not be the best solution for, as you said, worker safety, productivity, or efficiency.

MH: When the network is an afterthought, as you say how does the economics of the situation dictate the rollout?

JW: When you look at new technology like 5G, a lot of times people get frustrated at how long it takes to deploy some of these networks. And as we move higher in the spectrum frequencies, you need a denser network, and it is taking time for the public networks to catch up. Luckily, now we are working with Nokia quite a bit around private networks and that is just a different deployment pattern so that we can surgically fix some of these situations faster. But when you work with the economics of that, a lot of times we have run into, and we are already working with clients around the situation where you have an upgrade of the network that just happened for millions of dollars, and it could have been upgrading the IT Wi-Fi systems. And you have already gone through that spending. And you are supposed to be getting these benefits from it on the IT side, which you are.

And suddenly, they hear about a 5G private network or a private LTE solution. And they say, I do not really have an additional budget for this. I am not ready to now commit dollars to this across my entire landscape of 13 manufacturing plants, for example. We are running into situations where we have to one, find and prove the economics, two, start with a pilot to prove the benefits and the break even, and then three, slowly work into scaling it out. But a lot of that is because, with new technology, the current enterprise has already gone through an upgrade cycle. They are already tight on spending. We are running into a very difficult economic situation that has a lot of uncertainty about it. So, you really have to be very tight on how you are positioning this, how good your business case is. And then you can see the value.

MH: Right, you see enterprises spending 20 or 30 million upgrading their existing Wi-Fi networks, as you point out. And then suddenly there's this big, new, shiny thing that is required to make the use cases viable. How do we find the dollars to support spending even more money?

JW: Yes, and we found almost two different types of clients in the market right now, and that is not where it will stay. It is just where the bifurcation is happening initially. There is one group that has just a very, very forward-thinking technology group that says with private networks, with 5G, with these new wireless technologies, we absolutely need to take advantage of it. Let us find the budget and squeeze other parts of it out. Potentially it is savings from cloud. As they move off these data centers, they are getting these savings going to an as-a-service cloud model and they are repurposing, or we are helping them repurpose those savings to new wireless technologies to make it a platform like a private LTE or private 5G network. That is one group of clients.   

We are finding another group of clients that are exactly what you just said, Michael, there, they just spent 25 million. We have helped them modernize their Wi-Fi systems. And now they are saying what can we do to figure out cost savings and bring that break-even point in fast enough so that we can apply this to the rest of the deployments? So, we will do a pilot, we will do a business case, and we will say, we believe you can break even in this amount of time. With this many current use cases, we will test it out and come as close as we can to those. And then it gives the company the confidence to say, okay, now I can deploy this in 13 or 14 different plants or sites based on the business case you put forward. And my costs come from X to Y through Z timeframe.

MH: I am fascinated to learn that Accenture has built an entire business case system using artificial intelligence to come out with what the output needs to be.

JW: As we went through this journey with Nokia, we found certain blockers on the business strategy side of this. It was not necessarily the radio or the core that was the issue. It really became the actual soft things, like the actual funding or the talent. And when you look at the funding part of it, like we just described, we ended up having to build solutions around it so that we were not stuck in that pilot paralysis scenario. We built this. It is going through the US patent processes, patent pending right now, but it is a business model and a business case that takes the inputs of the different use cases, the different requirements of the network, for example, how many square feet, what type of radio, is it an on-prem core? What type of edge compute, all those things?

And as it takes all of that, it goes through a machine learning and AI solution where it spits out—this is when you can expect a break-even point—this is when you can expect the actual payback to happen. And dynamically, as you change and increase the number of use cases. Whereas you change the current requirements, the number of radios for density, it is automatically adjusted to it. That really gives the C-suite of the enterprise a lot of confidence that you have dealt with the sensitivity, you have dealt with the different scenario planning, and you would have a reasonable level of confidence to start this trial—please allow us to move forward.

MH: At the end of the day, the C-suite is interested in when you are going to break-even on this, when it is going to start being accretive to the business, tell me though about everything being sold as-a-service these days. Does it help the business case to go with a network as a service?

JW: There is a lot of discussion we are working with clients on in that scenario right now. We are seeing situations where clients will ask us to work on the old model where the client will purchase the hardware and then potentially for management—they will come to different groups to do the management of it. There are other scenarios where they are saying—just like we are dealing with the cloud buying dynamics—can we think about—and as-a-service model—where our CAPEX (Capital Expenses) is just tight right now, we are having a lot of difficulty unlocking spending, but our OPEX (operating expenses) is looking okay, so can you help us with an as a service model? But that has caused us as an industry to have to work together differently in partnerships, right? I mean, typically in a resale model, we have always been very used to reselling hardware or dealing with the warranties.

But in this model where they are taking that as-a-service model, we all must negotiate working together. The ecosystem must change how they deal with it, deliver what the client's looking for. But each industry is different. For example, the tradeoff is that there are certain critical infrastructure companies and clients that we work with that require not only owning the hardware for security reasons and control, but they also want the monitoring to happen onshore, nearshore. They also want this certain set of security requirements. There are different SLAs (service level agreements) that they require. It really depends on the industry that we are working with, but we are seeing more clients ask for an as-a service model.

MH: If use cases determine the needs of a network and it is build-out, does a network need to be purpose built? Do you have to scrap all the old infrastructure and rebuild from scratch?

JW: We're dealing with both greenfield and brownfield scenarios right now. Greenfield bringing like a new build or a new factory and brownfield being retrofitting. And certainly, it is a lot easier to deal with Greenfield scenarios, which we could design from the beginning. What is the actual family of use cases of the problems that we are trying to solve? Put it into that patented business case modeling tool that spits out the break evens and the actual benefits, and then start to go into traditional network design, deploy, and run. There are other scenarios where we are retrofitting and that is more challenging, right? I will give you an example. We are working with a large furniture manufacturer right now, one of the largest in the world. And they were already working with Accenture on other parts of the transformation and said, “Hey, we are hearing a lot about 5G. We are hearing a lot about private networks. Can you come talk to us about it? You are already working with us on other parts of the transformation.”

When we came in and started to discuss that, what we found is that when you are starting to deal with patchwork networks, like you talked about before, we purposely built our video analytics solution for worker safety on Wi-Fi to start. And we told the client this at first. We said we want to prove to you why cellular is vastly different than Wi-Fi. And they said, well we just spent this money upgrading our Wi-Fi just use it. Right? We built the solution and we said, okay, it is going to be a video analytic solution where Michael is on the line, and he is working in front of a drill press that's fast spinning. And we spend millions of dollars a year on worker safety and worker comp issues because your hand gets hit by the drill press. Your gloves get spun up by fast moving machinery. And some of the worst things can happen. Can you help us solve this?

So, we built a solution where the current video analytics determines how far Michael's hand is away from the drill press. Use case one. If you do not have your protective gear on like your hardhat, your eyeglasses, and things on, we just do not start the machine. You are not allowed to start the machine unless you are prepared. We avoid it. The second use case. We used to let us say Michael's hand, unfortunately, gets on the drill press by accident. And the drill press gets close. We will give you an audible warning. That is loud enough over all the different sounds on the machine floor that it can tell you your behavior is dangerous. The last one we built out was that if that drill press comes down and we get to a certain distance from your hand on the drill press, we stop the machine instantly. We just cut the power. So, we should avoid that massively catastrophic damage that could happen. 

Now, when we built that on Wi-Fi, we knew the latency would be higher. But what we did not think about is how big the standard deviation from those latency measurements was. We found that roundtrip latency was 300 milliseconds roundtrip from network to application and back. And the standard deviation could have been a 30 to 35 millisecond difference. You can get between 270 to 330 milliseconds difference in Wi-Fi. It is a big gap. When we deployed the same video analytics worker solution onto private cellular, we found that, of course, the latency came down, right? Roundtrip came down to a double-digit level, but the standard deviation was six milliseconds. It was incredibly more reliable on the private cellular network. You could really count on it when you are dealing with something as dangerous as saving lives. 

So, when we built it side by side, the client saw it and they saw the measurements and they said, okay, now I get this. Now I believed you before. That is why we started this program, but now I understand why you are saying there is a difference between the use cases you build on private networks on cellular networks versus the shiny new Wi-Fi system we already have in place. And now I clearly understand what you mean by an IT system that moves data around versus an OT system or network, that is monitoring critical events.

MH: The conversation about security systems to ensure that workers are safe, etc., has me thinking about different security systems, the securing of an OT network infinitely more complex in securing an IT network, when you have the risk of lateral threats from bad actors, moving from a camera to an AI system that is figuring out what that camera is seeing. The days of single ring fences are over. How does an enterprise avoid, the ‘whack-a-mole’ of trying to stop intrusions from skipping around a network?

JW: It is such a good question, right? I mean, if you just look at the example, we just talked about Michael, around worker safety and video cameras. You can imagine that now you have this increased threat surface; you have diverse types of devices. It is not just smartphones or tablets, which are already difficult to deal with in an enterprise network. Now you have cameras. Now you have sensors. Now you have machines. Now you have all sorts of devices out there. You have new network elements that you were not dealing with before, you might have an on-prem radio. You might have an on-prem compute or a core. Now you have this network that is different. You have an increased number of devices and types of devices. The threat surface is massive now. This ‘whack-a-mole’ concept you bring up, which is such a funny visual to me because it is so true. Before, it was okay to connect a circuit, put up a firewall and move on to the next branch site or the edge location.

And now that is just not... You just cannot do that anymore. Right? I mean, one, it would take forever, and it is not flexible enough, but two, you cannot deal with that threat surface. So, we must completely re-architect. What we are doing is working with clients on not only the policy and the philosophy and bringing zero trust networking into the very beginning of the design process, but we are also changing the architecture of the network so that we are making sure that these end points, these network elements are all being secured differently. And it is not just a putting up a firewall situation and hope for the best, but it is re-architecting the network and using more of a zero-trust policy-based solution.

MH: Right, so you must change the philosophy right up the chain of command. But I am fascinated that Accenture is involved in enterprise network decisions. I would have assumed that it is typically a conversation with a hardware vendor.

JW: And it is. What is different now, I think, is that there are a couple big trends that are happening. Number one, certainly there are more hardware vendors. Now that there's hardware and software desegregation, there's just new players involved. So that has become one big trend. The second big trend is that a lot of these enterprises are already going through big transformations in other parts of the business. We call it total enterprise transformation. And as you are going through that, Accenture's already involved in a lot of that total enterprise transformation, whether it is in the back office or the front office, whether it is in other departments, Accenture being a 700,000 plus person company. We are already helping much of the G 2000 with different parts of transformation.

As we get into those discussions, and as we get into those deployments and implementations and finish out the transformations, a lot of times this conversation about network transformation comes up, or we proactively mention the discussions around the need for a modern network that is flexible enough to keep up with the speed of business and how we can think about the network differently. And that is where we have a good partnership with Nokia, to have these conversations.

MH: On a previous episode of Real Conversation, we had Industry 4.0 founder, Henrik von Scheel tell me that people are at the center of transformation. What's Accenture's view on that?

JW: Massively important. A lot of times in the other transformations that we are working on, you will be able to really have a good conversation with the C-suite, who agreed to the need to transform, the need for the speed of this and the flexibility and the different cost savings that are coming up. But when you get to the next layer, when you start to try and do these projects, here is an example. You will find that the Senior VP of engineering grew up with all those engineers. He grew up with everybody from that department and you come in to talk about transformation that does affect those jobs. And suddenly, you run into this roadblock of what a great idea, the C-suite wants to do it, but then you find a thousand reasons underneath asking why. And a lot of it does come down to, well, are we going to get rid of all these engineers that we have grown up with or are we going to find a path forward?

So again, just like when it came to private networks, we had to find a solution, build out an AI-driven business case for transformation. We also have to start to look at a clear plan on talent and up-skilling and training and new plans for those folks that potentially may have jobs, for example, a worker-safety one. You could use the same video analytics solution on the same private network. And we have also repurposed that to quality assurance. In that quality assurance, let's say Michael and his colleague were checking the actual tolerances of a widget coming off a manufacturing line. Now what are Michael and his colleague going to do when we are using video analytics to check the tolerances of quality assurance?

Well, we are going to have Michael up-skill in a different part of the business, right? So, we trained Michael to potentially build algorithms, potentially checking the tolerances over the video analytics, or potentially doing other parts of the business. And that plan then gives confidence to the Senior VP level, who says—oh, you have looked at this holistically? —It is not just an automated solution on a private network that is more reliable—but you also have a plan for our people? —And you have an upscale plan and a new placement plan for each of our people in the business? That unlocks a lot of that challenge.

MH:I suppose since people are often resistant to change, we also must focus on up skilled change management. Do you get the sense that enterprise, the C-suite, recognizes this is a significant issue?

JW: We hope so. We are trying to make sure we are driving awareness of that, right. There are a couple of things that we are really pushing hard on with awareness. The first big message is that a modern network is required for digital transformation, and it is a competitive advantage. The second message is, if you buy into that thesis, that the modern network is this competitive advantage and required for digital transformation. And how do we find the actual talent and up skill plan, so that we consider everybody involved. And then the third part is what type of funding and sources of funding do we think about? And those are the three big messages we are trying to drive awareness of as we go through this journey.

<< Go to previous episode