Cisco Industrial IoT with Ruben Lobo

Hi, folks. Good afternoon. Um, my name is Ruben Lobo.

I'm going to be co-presenting with Andrea Oli, who's at the, uh, back of the room here. So we are from the, uh, industrial iot business unit at Cisco. And what I'm going to cover for the next 30 minutes is, uh, give you a little bit of a preview of what we launched and talk about some of the new use cases.

Uh, we are releasing, uh, through these new products that we launched. So, um, most of you know us as a, you know, a networking company, a security company, and you're very familiar with what we do, uh, in the campus and branch and in the data center and with service providers. But within the, uh, industrial business unit, what we cover is, uh, connectivity outside of your, uh, carpeted space, essentially right outside of the office.

And we, what this, uh, kind of chart is showing you here is ization in increased amount of complexity as it goes through maybe just connecting like a camera outside of a building down to like manufacturing factories, down to utilities, all the way to very rugged environments like mines and, and, uh, uh, uh, refineries and so on and so forth. And so what we focus on as part of this business unit really is to bring a consistent network architecture all the way from the enterprise down into like the physical world where we are enabling automation. That's what the focus is on.

And we take, uh, the technology that we've built within, uh, iOS, xe, you know, ICE Catalyst Center, all of those technologies, we bring it out into connecting, uh, the real world. So with that kind of context of what we do, um, just wanted to set, uh, you know, what we've been seeing over the years. Now, we've been in this space roughly for two decades now, right?

And we started with one switch down to like an entire portfolio that we offer both in, like in industrial switching, industrial routing, industrial wireless, and industrial security. Like, so we kind of, kind of, kind of like covered the gamut. And, and over the years, uh, we've grown to become the largest, uh, vendor in the space.

Like we've got the highest market share. And the big reason for that is that we are seeing an increased participation of it in operational environments. Like, and, and these days, you know, uh, covering OT has become like a priority for it, and it's mainly driven by two factors.

One is cybersecurity and AI readiness. That's the new trend that we are seeing in our talk. I really want to talk about like how we are enabling these use cases, uh, with the new portfolios that we've launched.

Um, and the key thing to make this possible is, uh, what we like to refer as it OT collaboration. I know in the industry the term is it OT convergence, but the reality is that IT networks and OT networks are never going to converge. They're, they're different lines of businesses.

They serve different use cases, but it's really important that these two teams collaborate in a, in order to kind of like digitize these environments. And, and we've seen success with people, with customers who have managed to like bridge this gap between the two, uh, organizations. So two more slides, and I'll get straight into like the launch material and get into the technical stuff.

So to set the context, last year we did a survey of our customers around, we had around a thousand respon, uh, responses. We asked them about various things, their top care abouts, their top challenges, you know, what are the trends in the industry. And around 48% said that they expect AI to cause like the biggest shift and the biggest impact in the environments.

Now to be clear, AI in these industrial environments has been around for many years, right? They've always done machine learning. They've always done like, uh, uh, uh, neural networks and, you know, applied those technologies.

Really what they're talking about here when they say biggest impact is the use of generative AI in these environments, which is kind of like new for them. So with that context, what are the use cases that we are seeing among our customers? So I've listed a few here, all the way from machine vision to, uh, autonomous vehicles down to, uh, robotics and edge to cloud.

And the key point I wanted to drive here is that every one of these use cases has an implication in the network. If you look at the adoption of AI in industries, machine vision by far is the leading use case that you see across the board. Doesn't matter which industrial vertical you're looking at, vision systems are at the, at the leading edge of it, you know, the adoption of ai, and they're causing an increase in the amount of POE ports that are being consumed.

They're driving up the bandwidth on these networks. We've gone from customers, you know, mainly asking for fast ethernet ports, which most of the time a drive or an IO device or A PLC generally has, you know, a hundred meg down to like, to a rapid evolution to POE and, and one gig ports because they need to power these vision systems. And that's also driving bandwidth in the network.

So now we've gotten this request for, uh, for, for 10 gig on the uplinks. Um, we've got a lot of autonomous vehicles being adopted in factories, um, in, uh, food and beverage, um, in, in data centers as an example where you need ultra reliable wireless for these AGVs and AMRs to move where wifi doesn't cut it, right? And then you've got this move to virtual PLCs across the industry is known as software defined automation where control is getting detached from a physical device, which is A PLC and getting virtualized on general purpose compute.

And there's implications of what the network has to do in order to support this, uh, this kind of use case. Um, and then, oh, talking about AI robotics, we see the same thing happening. You've got, uh, robot controllers that typically is a big kind of, uh, uh, uh, CPU kind of device, a pc let's say, sitting right next to the robot.

And it's very constrained in what it can do. And what we see the trend in the industry and what customers are doing is to take that robot controller, virtualize it, put it in the data center where there is a lot of CPU and GPU available so that they can program that robot to do complex tasks. And finally, on the edge to cloud use case, I'm showing an example of a traffic intersection here where you have cameras detecting passenger pedestrians going across, uh, the crosswalk, let's say, or an accident on the road, uh, inferencing happening on the edge, and then communicating with an orchestration system that sits in the cloud that is, uh, orchestrating the traffic or changing the traffic pattern across all the traffic intersections that you have, right?

So these are like some of the leading use cases that we see. And in that particular case, you need assurance between the edge up to the cloud through owned and unknown networks as an example, right? So that, that's just to set up what we are seeing here.

So what we believe that will happen in the future, not tomorrow, not in a, maybe in a couple of years, we can't tell you if it's like a few years, five years down the line, let's say, uh, is that we expect the brains, which is the control, the virtue of the PLCs, the robot controllers to slowly get decoupled, uh, away from, uh, physical devices, right? And, uh, move up to the data center. So you're still going to have your io your drives, your robots on the factory floor, but eventually we start, and this trend is already starting.

You start to see this control getting centralized in the data center where you can train, uh, uh, train the, uh, the software with models and do inferencing at the edge. And to make all of this possible, the network is key. Without the network, you'll never be able to realize this use case, right?

So that's, that's where we see this industry going. It's not gonna happen tomorrow, but there's clear science that this is what's gonna happen. And within Cisco and within industrial networking, we're very, very confident that this is, this is the trend.

So with that, let's just dive into what we are launching here. So we've got three big announcements that we're making at Cisco life. The first one is a brand new portfolio of industrial switches that are going to enable a lot of the use cases that I just spoke about, and I'll dive into details and explain how exactly we enable those use cases.

We are redefining wireless support for the industrial use cases. We are bringing wifi and our ultra reliable, uh, technology closer together in, in one single access point using one management system, and Andrea's going to cover that. And then last, but not the least, is when it comes to security.

This is an example where we are using, uh, native AI in our products to make the lives of our, our users simpler. The first two are examples of where our infrastructure is enabling AI use cases for our customers. The last one is an example of where we are using clustering to make it simple for our users to be able to group the assets.

That cyber vision, which is one of our products, uh, helps them discover to auto group those assets, uh, in order to build segmentation. Like a lot of these networks are super flat. Segmentation is a challenge, and we help them do that, and I'll dig into that.

Okay? Any kind of questions before I, I, I dive in good setup. You kind of understand like what, where we, where we are, what we're doing.

Um, I'll walk through switching, I'll give it to Andrea to cover wireless, and then I'll come back and close it out with security, okay? And then I'll give you a, if you've not been to the world of solutions, I'll give you a taste of what you can expect and, and make a, a plea for you to come there because it's definitely gonna be worth your time. Okay?

So there's a lot of use cases to cover, but we have very little time. So I wanted to take two use cases and kind of show you what the implication of these two use cases on the network vision systems and virtual PLC, right? With, if you look at each of these vision, uh, each of these use cases independently, you miss the point as to how important the network is, uh, is to make these use cases possible because they have kind of conflicting requirements that kind of like make it difficult to deploy the two use cases when you deploy them in parallel, right?

So let's dig into what this looks like when it comes to vision systems. What we are seeing is the evolution of vision systems of machine vision towards a network, uh, machine vision. So on the left hand side is what a typical vision system looks like in manufacturing today.

It's, you know, if you, if you're familiar with this, this field, they're rule based, right? So there's a set of rules. You program these vision systems a according to a set of parameters, and they're able to do things like find defects in the products as things are getting manufactured.

As an example, the key thing to notice here is that they typically tend to be siloed. There's an IPC that's used that runs the, the program for the vision system, and the cameras may be PE powered or maybe not, right? Sometimes you power them with IV.

But the key point is they're like this isolated, siloed systems programmed one at a time. They do a good job, but it's, you know, it's cumbersome because you gotta kind of like, uh, you know, tackle each use case by itself. What's happening?

All the camera vendors are moving towards what's happening on the right hand side, which is they're now starting to use, um, um, AI to be able to train the model, train the model based on what the, uh, vision system is, is, uh, is, is, is seeing at that point in time and then run inferencing on the edge to be able to detect a lot more complex things, uh, um, uh, in, in terms of, uh, being able to detect defects and and so on. And there's several use cases that they enable. But the point here is that now what's happening is the cameras are purely powered.

That's kind of like table stakes, but there's a lot of data being sent up into like the level three of the podium model or into the factory server room where the models are getting trained or the images are getting stored for, you know, uh, defect tracking in the future and so on and so forth. And so it's really driving up bandwidth on the network. Like this is like a big trend.

We've gone from fast ethernet to like 10 gig couplings all of a sudden, and it's happened in a matter of like two years, right? This is a big shift that we are seeing in the industry. The second thing is the emergence of the virtual PLC.

This is the use case where we are actually decoupling control from A PLC and putting it on general purpose compute for the purpose of, uh, driving efficiency and automation. So the automation vendors are focusing on how to enable deploying software, deploying PLC programs and maintaining these pro PLC programs on, uh, the plant floor, just like how it deploys software, um, in the, in, in, in the enterprise, right? Like that's where the, uh, the, uh, industry is shifting.

And so we have two kind of like ways in which this is playing out. The PLC is getting detached from dedicated hardware. In the first case, it's going up and getting centralized in on servers in the data center.

If you're, if you're not familiar with the use case, there's a big announcement that Audi made with Cisco and Siemens where they actually took their controllers, virtualized it in the data center, and it was the first time it was ever done. It was like kind of like groundbreaking breaking in terms of what they did. Um, and, and, and we helped them enabled this use case.

Um, the key point here is that the control traffic, which is super sensitive to jitter, is now sharing the entire factory network. On the right hand side is what we like to call as automation in a box where the virtual PLC is running on regularized, uh, PCs on the floor, but it's sharing the, the CPU with other workloads, right? And now this PLC is sharing traffic with other, other applications and network.

And so all of a sudden what you have is deterministic latency that's required, uh, for these virtual PLCs that is competing with the video in the previous use case, right? So you have big jumbo frames of video traffic, and you've got super latency sensitive, uh, jitter sensitive control traffic that all has to share the same link. So introducing the IE 3,500, uh, switch that we just launched today at Cisco Live.

What does it do to enable these use cases, right? I'm just picking a set of features that this particular switch, this industrial switch does to support these use cases. First and foremost, we have really amped up the P OE budget on the switch, the number of POE ports, uh, to support, uh, vision systems.

5 gig, uh, down links to, uh, uh, support, uh, some of these pan and tail cameras. We have 90 watts of POE. You can connect more POE cameras to the system than we could in the existing, uh, shipping product.

We've got things like fast, p oe, perpetual, POE, all of these things to help, uh, provide resiliency to the camera so that when, uh, the switch kind of, let's say reboots, when you're doing a software upgrade, uh, the power still goes to the camera, or let's say you are, uh, or, um, let's say you have some planned downtime and you reboot the switch, reboots the camera gets powered before the switch even comes up, right? And then we've got three 10 gig up links to enable these video ca use cases that I just spoke about. Now, on the flip side, to support this deterministic traffic we've got frame preemption.

Any anybody's familiar with what frame preemption looks like? So if you, if you think about typical qs, right? You've got your priority queue, uh, but even if your high priority traffic is on your high priority queue, if a low priority frame is already on the wire, you have no choice, right?

You gotta let you know 1500 bytes, uh, uh, get onto the wire and then your high, your, your, uh, high priority traffic can, can continue. What ts and frame preemption does is you've got this notion of an express queue, and the express queue will preempt the low priority traffic. So let's say you've got two bytes of a 1500 by packet on the wire.

It'll stop that frame, let the high priority traffic go, and then resume the low priority frame and the other side reassembles it, right? And with that, we minimize the jitter for high priority, uh, control traffic, right? So we are able to support these two kind of like use cases together.

There's a lot of synchronization that takes place with robotics and so on and so forth in the environment. And we support PTP, uh, with boundary clock, transparent clock and, uh, and gram master clock so that these robots and drives can be synchronized with these cameras, um, as all of these things are moving. So just one example of what this technology does.

Okay? So I'm going to hand it over to Andrea and we'll get to wireless and then, uh, come back to, uh, security. Uh, so echoing what, um, Ruben was saying, uh, you know, there's different use cases where, you know, we see wireless as, uh, a, uh, a key enabler.

You know, we go from, uh, you know, manufacturing where you can see, you know, automated guided vehicles that require reliable, uh, you know, wireless communication. And then we see, uh, trends in, uh, uh, heavy industries where they're moving personnel from, uh, uh, the field and they operate their machines from an office or, uh, all the modern subway system, uh, uh, are built, uh, or light rail system are built, uh, without a conductor, you know, so they're driverless trains, and so they're require reliable communication system to actually safely, uh, move them. Or, you know, wireless is a key enabler in places where it's really hard or impossible to bring copper and fiber.

And, uh, just going into one example, you know, we see, you know, a manufacturing, you know, assembly line going from con convey belts, you know, to, you know, the free moving a GV, uh, system so that, uh, they can reconfigure the assembly lines in a much faster way. And, uh, and clearly to go from one, one paradigm to the other, you know, you requires a shift in technology as well. And, and so just looking at a, uh, one example here, you know, we have a, a factory floor.

You have your wifi seven access pointing, you know, on the wall, and you provide connectivity to personnel and, and assets. Everything works and everybody's happy. But, uh, what if, uh, you know, the environment changes, then you have a, you know, industrial shelving there, you have more robots, more, more assets, and now your signal doesn't go everywhere.

You have drop, uh, uh, drops and, and, uh, communication, um, drops happen and, and your operation. And so the easy fix for that is really over provisioning, uh, the network. And so you start deploying more access point everywhere.

So your network goes everywhere. You have a pervasive, uh, wifi umbrella, but your personal might be happy, some asset might be happy, but then you still have some of the asset that need to move from one place to the other in your, on your factory floor and wifi, uh, even the latest standard doesn't play well with fast roaming. And so you might end up in communication fault.

And so to solve these use cases, uh, you know, Cisco has a technology called ultra reliable wireless backhoe. And that really address two topologies. One is extending communication, uh, in places where it's hard to bring fiber and copper, you know, and, uh, uh, you extend the entire core network, you know, in on a traffic light where you need to connect, uh, you know, uh, sensors or cameras, uh, or on the other side, you know, the, uh, mobility, uh, uh, topology where you are assets that are actually moving in an environment.

And then you need a seamless, uh, connectivity, seamless roaming between when an access point and the, and the other to enable industrial automation systems. And it's really based on three these pillars, building an end-to-end overlay to support the data, uh, traffic and, uh, and support the fast roaming, uh, make before break and off. Meaning the client device that sits on the A GV, for example, you know, and negotiate the connection to the next access point before breaking the communication where it's connected.

Uh, while, um, uh, wifi, it's really a break me for make, you know, uh, type of roaming. And, uh, we have also a technology called multi-path operation that lets you protect the specific traffic. For example, the PC for automation systems traffic, um, over the wireless, uh, network are duplicating those, uh, packets over two uncorrelated channels.

So you have, uh, if one channel gets interfered or you're moving and you don't have that connection with that access point anymore available, you know, you still have the secondary path that can, uh, uh, still send, uh, the information across the network. And what we are announcing, uh, you know, uh, as Cisco live, uh, uh, now it's really integrating these, uh, two technologies together with one single management platform, one single, uh, wireless infrastructure. And, uh, and so you can get the best of wifi performance for your, uh, IT use cases, and you can combine them or stretch your, you know, platform over, uh, and cover all the different use, uh, OT use cases where you are have to connect, uh, assets that require, uh, ultra reliability.

And so while you have, you know, to until yesterday, you know, what our customer needed to deploy is a dedicated IT network, you know, for their, you know, laptop computers and personnel. And then they needed to invest in several different, uh, dedicated industrial technologies to cover the ot, um, ot, uh, use cases. Now, instead of using the same platform, you know, same technology, you know, from the same controller, they can actually deploy order access points and decide which radios slot, you know, will be dedicated to, you know, wifi or which one will be dedicated to the ultra reliable, uh, wireless use cases, and connecting, uh, moving assets or extending that connection, uh, over in places where it's hard to bring, uh, network drops.

We're targeting, uh, a list of as use, you know, so we're clearly adding this functionality in, uh, the wifi seven access point. We're targeting, you know, the wifi six C and all the industrial portfolio, and then we're backporting that on wifi six as well, because you know, all the customers that are already invested in, uh, in Cisco access point, and now they can go into the WC and enable this functionality and deploy that, um, in their systems. When we tested, you know, in the technology out in the field, this is, uh, one of the major, you know, uh, automotive manufacturer that deployed our technology to, uh, you know, um, control their AGVs.

You know, you see the delta between, you know, a wifi six device, you know, that, uh, you know, clearly create, you know, delivers a very good, very good performance, you know, but you have a completely different, uh, uh, you know, situation where you deliver the reliable wireless back or, you know, you have a sub milliseconds, you know, latency you have or, uh, close to zero, zero packet loss, you know, on the network. And that is the enabler for, uh, pushing industrial automation systems. So just to wrap up, you know, uh, with this integration of the two technologies, you know, you can run both at the same time, uh, off the same access point, uh, through a unified, uh, uh, image.

And, uh, uh, that eliminates the need of deploying a separate different networks. There is a single unified management that can allow you to operate both and, uh, um, and, and so position Cisco in a very unique, uh, uh, uh, way that, uh, can deliver wifi and, uh, wireless connectivity, indoor, outdoor, and all the way to the, um, most challenging industrial use cases. Ruben, if you want to come back and cover, Thank you there.

Okay, so, uh, I introduced you to the, uh, IE 3,500, right? So, but we are not just launching this one switch, we're launching an entire portfolio. Like I said, we're introducing 19 new SKUs into the portfolio.

So we've got six SKUs of the 3,500, we've got the same SKU in the IP 67 form factor where you can literally like put them in food and ware and manufacturing and essentially wash down the switch, as an example. We have it in the compact form factor. And then we have the I 30, uh, 3,100, uh, offering four PPPE.

So, you know, imagine an enterprise use case where you've got this, uh, particular switch, and then you are deploying outdoor to connect access points or to connect, uh, uh, cameras for video surveillance, right? And all of this, uh, the switch is starting with the 3,500. Eventually it'll make it down.

Its portfolio will be managed by the Meraki dashboard. So you'll have one unified management plane for the entire kind of like product line. And the last switch that we're launching is the, uh, rack month switch with both copper and fiber ports.

Uh, with that, I know we are running towards the, uh, short of time, but I just kind of wanted to show you one thing. So this year we had, we have like, probably the most spectacular setup we've set up in the world of solutions in terms of display. So everything that Andrea and I spoke, uh, uh, shared with you, you can see in display in each of these use cases.

So this is exactly what's happening on the world of solutions, uh, today, right now. And this particular demo, which is a factory demo, is really, really impressive, where you can see machine vision and virtual PLCs in action. It's a super big demo.

You can see guitars moving and you can see robots tuning the guitar. So please make the time to come to the world of solutions if you can. And if you do show up and you reach out to us, we'll organize a special tour for you.

Maybe if you want to take some, you know, videos and put it on, on social media, we'll be happy to walk you, uh, through the setup and kind of like give you a quick, uh, uh, preview of what this looks like. Okay?