Expert Learning Session | The Role of Edge Computing in Driving Digital | Schneider Electric

Expert Learning Session | The Role of Edge Computing in Driving Digital | Schneider Electric

Show Video

Hi everyone my name is dalia dieben from stl partners. Today i will be, talking, about the role of edge computing, in driving, enterprise. Digital transformation. Um so we talk about edge computing it's always important to do a quick definition, of what we mean. Um, i suppose one thing is to put into context, and from an enterprise perspective, it's really about, changing, the way. It. Infrastructure. Works and ot infrastructure, works. And. Providing, enterprises, with more options so before, as an enterprise or a developer. There were only if you know at the moment even there are only a few options, of where, to, uh run and compute, and store data, and that's usually either on premises, on a in a data center. Or it's in a remote data center or the cloud. Or sometimes you know a lot of that data is also processed, in the end device which could be, in a car it could be in a machine in a factory. Um. You know, on a laptop on a mobile device. And what edge computing is all about is moving. That compute, and processing. In particular, and sometimes storage. Closer to the end device, or closer to the, the source of where the data is generated. Um so education, is partly a physical aspect it's really about you know. The computing, physically. Closer, to that source of data, but it's also a software. Aspect in the sense that, it's really about. Enabling. Enterprises, to be able to move those workloads, easily, and to be able to access. Um, infrastructure, in a very flexible, and seamless, way. Um. So the, the question is always you know then what are the benefits. And, there's two ways to look at it, um one thing is to say okay well my, sort of my status quo my default option is to host. Those, the workloads, in the cloud. Um and so you know why should i move, the the cloud, to the edge or you know why should i move those workloads, to the edge. And the first reason, is, latency, it's always about latency, that that's what tends to be brought up in these edge compute discussions.

Um And low latency. You know is you know due to the fact that your, the data pack is having to to travel. Um across a, sort of a shorter distance, if you will, um that reduces, the latency. Of the application, and so things can. The round-trip, time and the overall. Application, latency can be reduced, dramatically. By reducing. Um. You know the time between. The data. Being sort of produced, and then processed, and sort of taken to the cloud. The second thing is, um, is the cost of bandwidth, and, again you know there is a cost associated, with moving those data packets, through, network or through multiple networks if we're going to the cloud. And bringing that, uh that compute resource closer avoids the need to have to traverse, all over. Um you know the full network. The last thing is data security. And this is becoming. More and more prevalent, and more of a, sort of a key priority. For enterprises. Partly because of regulation. And you have. Increasing, regulation, that's being. Done at a sort of a national, level. Um. And also sometimes it's even at a sort of a state or municipal, level. And that can make it very complicated, for an enterprise to manage data. And also to ensure that data restored. Especially. Um. In a sort of federated, way, and, um. The enterprise is aware, of where the data is stored because you know there are compliance, issues related, to um to data storage and security. And the other thing is that, you know enterprises, themselves. Are conscious. Of their own, privacy, and security, and you know cyber security, issues. Um, and. You know we've heard a lot when we've done interviews, say from the manufacturing, industry. Um, there are, the manufacturers, themselves. Are very, sensitive, about their data their sort of precious ip. Being in the hands of anyone else so if they're running an iot application, in the cloud, for them that's a that you know that could be a sort of a security, risk for them because, there is a cloud infrastructure, company, or. Even an iot. Solution provider, who has potentially, access, to, you know all the ip related to how their manufacturing, process works. Then the other way of looking at the benefits of edge is saying okay well. A lot of my computer actually happens on the device, or maybe in sort of more monolithic. Um, architectures. On premises, or in sort of traditional. More dedicated, appliances, where the software is tightly integrated, to the hardware. And there isn't this sort of cloud-like. Remote, management capability. And again you know the reasons there for moving that to an edge, um.

First Of all it's to. Reduce, the cost, of infrastructure. Where. You have, um. Where you have you know a lot of the time an enterprise has to invest quite heavy amounts of capex. Um to put an on-premise data center in place or you know if if each application. You know each, software, is on a separate piece of hardware, that adds up and that takes up space and also, it's costly. Um and you can see that in a certain environment say in retail, that's, particularly an issue where. They, that real estate is expensive. And they don't really want to be using. Um, that press sort of precious space, for. Um putting in servers they want to be able to consolidate, that and so taking that off the premises, has huge benefits. Um the other thing is mobility, so, the i mean the challenge of putting some tying something to a device or tying something to a physical location. On premises, is that you then can't access that application. If you know if you. Move around. And you need that level of mobility. And. You know enterprises, i guess have, um. Have been able to. Um access ic infrastructure. To. Enable. Branches. To access applications. But it becomes more difficult when you have a very remote workforce, where, you know people are working from home or people are, um. Or you know there are. Workers, sort of you know in the logistics, field or field force workers, they always need to access the applications. But also need to, access, them in a way that's, resilient, and you know with low latency, and that's why, you may want to. Run those applications, at the edge. And then, the other sort of the last. Two points, are about. Uh, increasing, flexibility, and scalability. So. Um again you know bearing in mind that edge compute is also, sort of about edge cloud if you will or a software element, and it's about bringing those capabilities, that we have in the cloud around flexibility, and scalability. To. Um. Compute, resources, which have traditionally, been quite. Siloed, in nature, or monolithic. Or not as flexible. And, um. You know by having sort of distributed, com, cloud infrastructure, almost, on.

Which Could be on, gateways. Which could be in servers. Which could be in, many data centers. Being able to access all that compute resource, in a cloud-like, way, um, is, beneficial, because, it again reduces, that upfront, capex, it makes things it's, more agile for the enterprise, where they can scale, up. Demand for their compute resources, when they need to. And then the last thing is that you know by by sort of disaggregating. Software and hardware. And, um. Making. Edge compute, sort of uh accessing edge computing and, using, standardized, flexible, tools. That reduces vendor lock-in, where you don't have a single vendor to provide you the application. The software, platforms, and the hardware all integrated. You can go to different vendors for each part. And this and this sort of next slide just, shows that in a different way, um. Where you have. Um. I guess this framework that i've just described. Um, in a nutshell. So. You know just to reiterate. Low latency, reduce back call data localization. Those are key, characteristics. Of computing. Applications, locally, historically, so, an enterprise will always, put their applications, on premise, if if it's mission critical and low latency is critical. Versus. Cloud like. Capabilities, are associated. With you know the scalability, flexibility. Aspect, mobility, so being able to access the applications, everywhere. Um. Having that level of resilience, which the cloud brings that sort of thing and so the edge is really bringing those two sides. Together. And these benefits. Um, differ. Across each different industries and this is quite a high level view of looking at it but, you know you as i mentioned you know we've got. Industries like manufacturing. Where, much of their. The types of applications, which are going to be running on edges. Are it but also the ot applications, which are mission critical, to running their factory. And so, things around low latency. Reliability. And keeping data local, are paramount, to them, but so is reducing back core because, by, with all that iot, data now being produced. Um with hundreds of cents in a manufacturing, plant. That is, unfeasible. To be um to sort of go to the cloud because of the amount of raw data, that it will cost. Um, to, try you know to go over the network. You know on the other hand you have i guess um. Industries like retail i mentioned, where, there it's more around, maybe, taking, applications. Or. Accessing, applications, which may have resided on a single device before. Um, so for example, and we'll go through a few a few, real life examples, here but, in you know in the retail, store, you are trying to create immersive experiences. And, a lot of the time that is tied to a physical device which could be a big screen a sort of digital screen or it might be an augmented reality, mirror. And, um. Historically. A lot of the computer has resided, on the actual device, and that's. You know that's not. Great for the for the, retailer, because. Um it's it's a capex, it's a huge investment that they have to make on those, expensive, devices. There's a security, risk to have, expensive devices, in their store. Um, and also. Um. Then they're inflexible, so when they want to upgrade the application, or change the experience that they're trying to create. That's it's difficult for them because they've already made that investment, in those devices. Um. And so, so yeah so broadly speaking edges, somewhere, between. Um, these. Devices, where software is tightly coupled with the hardware. And the internet, or where you know where the data centers are today. Um, and you know there are different almost levels of edge computing, where. You have, um. You have say the device edge, if we start there, um, which is, you know sometimes you can think of smart devices being a device, edge and, and this is a bit of a gray area, you know when is an edge device, versus just an ordinary, end device with some compute processing.

I Think the key thing is that. That. That um. That device, needs to be you know virtualized, if you will it needs to have, some of the capabilities, that you'd associate, with almost like a cloud. Application. And the types of applications, that should be deployed there you know it should almost be drag and drop where, it's a cloud native application, that you could you could run in, a remote data center, you can now run on a device edge. Um and so examples of devices here could be like a smart camera, where, um. The oems, are, making their. Um, these, devices, which you used to just, pretty much collect data and do some pre-processing. Allow them to do more of the analytics. And, um you know things like object recognition, that sort of thing. Um, and you know and i guess you know there's some quotes here we've spoken to a few, i guess companies, in the past, who are either solution providers. So here we've got video analytics solution provider and what they you know why they see there is a move towards device edge, and then another case we've spoken to sort of end enterprises. The second category, of edges. On-premises. Edge which is putting the edge compute, resource. Um. You know at this at the customer site at the enterprise site which could be a building, could be a um. It could be a factory. It could be a retail, store as i mentioned it could be a hospital. Um. And, a lot of the time. Um. You know this is, a lot of you know a lot of time this is about. Keeping, things, um, secure. In the. In the customer, premises. Reducing, latency. By, um, putting those. Applications. Closer to the source of the data. And then the last kind of category, is network edge or multi-access. Edge computing and this is coming from the telecoms, industry, where, they are evolving, their networking. Sites, so they have points of presence, or facilities, that are running their network. Functions, and their network infrastructure.

They're, Evolving those to become more like data centers. Um, and so, they. Um, by, you know by hosting applications, on those types of on that type of edge you dramatically, reduce latency. Because you reduce the need for the application, to go through the network and back and. Sort of reduce that round trip time. And, it's interesting, i guess to see that what the public cloud providers or the hyperscale cloud providers are doing in this space. Because, um they, are moving to the edge they you know that their bread and butter is the centralized, cloud. And, you know they did their data centers would be that sort of internet book as i had in the previous slide. Um, but you know i guess initially, they've you know they all have sort of some level of cdn, offerings. Um and they have a they have. Um, access, to more regional data centers, and space, in internet exchange providers for example. Um, and they are increasing their footprint in those data centers, but that's always primarily, been for their cdn, service. Or for their internal. Applications, if you will so for example. Even though google's not on here google. Uses its own, um, cdn, network. Um. And sort of cdn data centers for youtube for example to cache youtube. I think what's changing, is that they're now starting to evolve, that um. Those, uh, those cdn. Data centers. Or, the, premises, which they use the third part the colo data centers that they are using. And the stack that they put in there before it was all about storage and caching, now they are evolving, it to make it about compute, and so they will start to allow. Developers, to access, those, more regional, data centers and regional locations. And so aws, for example announced local zones which is a flavor of that and then zero has azure edge zones which is their sort of version of a, kind of a regional data center play. So as a developer. You could now go on to aws, or azure's portal, instead of clicking, um you know the europe south zone for example, you would be able to specifically, select. Um, a, you know a region within that, um so just you know hypothetically, it could be something like milan, or something. Um, and then, uh then. I guess the next sort of chronologically, what happened next is that they moved into the on-premise, or the device world and tried to bring in some iot. Um, services. With things like azure iot, edge, and green grass, and that was a software-based, solution. What they've done since has evolved that to actually provide a computer. Device, as well as a physical, element, and that's where outpost, and as your stack and azure stack edge come along. And then more recently, they're starting to move into the network edge and doing deals with the telecoms, operators, themselves. So that they can put in their stack. Which is essentially you know a rack. Of, um, of their own servers. Into the uh the telcos. Edge. And provide, their, their cloud services from there as well and so with aws, it's wavelength, it's their version of the network, edge offering, whether it's your they have something called as your edge loans with carrier. So go i guess going back to then you know the the enterprise, though and why does the enterprise. You know, where does edge fit into their digital transformation. Um. I mean we've we've kind of talked about the benefits of edge but it all really fits into. A. More, broad. Theme around digital transformation, because that's primarily, what these enterprises are trying to do they're trying to transform, so that they can either improve efficiencies. Whether it's about reducing, waste and defects. By. Automatically. Detecting. Faults. In real time, using sort of iot, devices. Or it could be about increasing, the asset lifetime, by doing things like predictive maintenance, to keep the asset running for longer.

So These are the types of applications. That edge enables. Which then enable digital transformation. And you can see some um you know some quotes here about. Um. Sort of, you know the dynamic between operational, technology, and i.t technology, and that's coming together that's a bit of a challenge but it's also, quite key, in edge computing. Um. The other side of digital transformation. Are these new business models so, so some of the underlying, objectives, are about improving efficiencies, and primarily. Saving, costs, but on the other side you also have, sort of the new business models element, where, in many industries, they're moving towards more service, type business models so it's going to, x as a service, or, um, taking manufacturing, as an example. Um you know they're doing things like providing. Assets as a service. Where. You know rather than, a traditional, business model where they will ask the customer to just pay for the, product, up front. Um they'll provide it as a service, and they'll. They'll bundle in um you know maintenance, services, and insight, services. Um, and allow the customer to know you know if it was an engine for example. That would allow the customer to know where the engine is how it's running. Um. You know, how if there's any problems with it that sort of thing. And that's, quite cool for them to survive, and to evolve and to keep to you know keep growing their revenue base. The second thing is just sort of a more data-driven. Business model, where, you have, um. The sort of data-driven, product development going on, and again you know just taking manufacturing, as an example, here. Um, they, you know they need, insights, on how their products are being used to be able to. Create the new version of their products, whereas before, they sort of sold them and they had no idea how the customer is using them who is using them why they were using them and all that sort of thing, and that's quite key for them to you know again for com you know for competitive, reasons to develop products that are, um. Beneficial. And now i'm just i'm just going to go through a few examples, here so. First one is in manufacturing, and i think i've explained some of this where, you have kind of an edge compute platform in the manufacturing. Plant, and it's doing things like transforming, data, aggregating, data from multiple machines. Filtering, data that's not you know that's not interesting, where for example. You have all this raw data coming in from the sensors, what you really want to do is be able to, filter out the anomalies, for example, and just send those, the spikes or the troughs. To a cloud. Some machine learning algorithms, might be executed, on the edge compute as well. Um and that enables, applications, like condition based, monitoring, was which is sort of what i mentioned earlier about being able to remotely, monitor. Assets. Predictive, maintenance, which is all kind of a step above that using, now ai and ml. On that data, to be able to predict.

When A machine is going to break down and you know. Make sure it's fixed ahead of time and then you have things like just general, real-time, automation, of the plant. And this is, you know fundamental, for manufacturing, as they try and, evolve. Their. Processes, to become, industry 4.0, and iot, based, and i've just got a few um logos, here to show some of the, some companies, who are providing some of these edge analytics. Capabilities. These sort of software platforms, if you will. Another big. Application, for edge, is. Ar and vr and sort of immersive, experiences. And here, i'm just showing, i guess different compute resources, where you have device, on-prem, edge network edge in the cloud. And what happens today is that. You have sort of, ar, applications, that could be um, rendered. So the the actual virtual reality, could be rendered from the cloud, and the problem with that is, you know latency. Um and it's very critical for ar and vr. Especially for virtual reality. Because, you. Your frame of reference is what you're seeing in front of you in that virtual reality headset so if you move. And the image doesn't move with you that actually. Sort of creates a motion sickness, element, and so, you know the milliseconds, that are required are 700, milliseconds. Between. You know round trip. Um to enable that ar vr application to work, properly, and i think what we've seen today is that actually. Um, most. Vr applications, are run from the device themselves, you have, the content loaded on the device or you have it you know the device, tethered, to a local compute. Or a laptop. To to do the actual rendering, and that's you know that's not ideal and that's not going to. Allow arvr, to scale you need it to be fully mobile and you need it to be sort of wireless. And so, um, you know, that's where, the network, edge or an on-prem edge can really benefit where you run, the arvr. Application, or you render it from a network edge so that you redo you can reduce that latency, to under 100 milliseconds. And you, um you then only use the vr headset, to display. What was rendered, from the network edge. Um and these you know these types of things are quite cool, in, um. In say a retail setting. Um, because. Retailers, are trying to as i said create sort of immersive, experiences. Um both, you know in at their stores especially in flagship, stores you have you know lots of, um sort of immersive.

Boards, Or digital. Signs, um you have arvr, happening. Um and also in the home where, you know. Today, e-commerce, is you know is evolving quite quickly, you can use your mobile phone to. Um, to sort of look at the product and maybe, sort of look at look at it on you virtually, or look at how a piece of furniture, looks in your room and that all uses augmented reality. In the future we may start to, you know we will start to see more of that and to have you know very high sort of high quality high definition, arvr. And as the, um. The quality, of the, the, um, the image in increases, or improves. And that is a huge strain on bandwidth and that again you know it. It drives a need to move that data not. Not run it from the card but, at an edge. And here we've just got a few examples. Of companies, again who are um, who are starting to to do things around this space, you know some of his arvr, some of it in the retail setting is more about, immersive, experiences, more broadly and using things like iot, and digitization. Technology. To be able to create. Um what what mommy, is doing which is a sort of augmented reality, mirror, so you can go to a changing, room, and, um. And you know. Try different things on virtually. Um, without having to actually get the clothes which is you know, quite an interesting content. Concept. And again just sort of bringing it to life so how is edge compute being deployed well there's, both brownfield, and greenfield, deployments, you have, um. Uh these these are sort of edge compute use cases like condition monitoring which i mentioned about remote monitoring. An asset, some of that edge computing, is happening, on a virtualized, programmable. Logic controller, so actually changing what that plc, does, and sometimes it's on a sort of a. Separate, device, a server or a gateway, or it could be you know deployed, on a rack of. Servers and premises, or it could be on a, kind of on-premise data center so every you know all these things are very customer dependent there are different ways of deploying, edge depending, on, what the needs are and whether there are other edge applications, which need to be deployed. Too. In terms of where we are so we have edge compute which has kind of happened. For some time, um. You know this is, sort of the the broad idea of on-premise, data centers is not necessarily, new. Um but. You know historically, it's been a bit more standalone, where you have the cloud, and then you have on premise and there's less of a distributed, compute spectrum. This idea of edge cloud is really about moving towards. Um distributed, compute or distributed, card more multi-cloud. Um. Also having sort of hundreds of different compute infrastructure, out there where, as we saw on the previous page these could be small devices. Or a single plc, or a single gateway, to, a whole rack of servers, or over, you know a full, data center. And, um, the software is quite important in edge card where it needs to be a sort of containerized. Environment. In particular, to be able to move workloads, quite seamlessly, across different clouds. And to be able to access those, um as a service business model so even the hardware, vendors are moving. Um to to. Sort of as a service, business models, um. And they can only do that if their, platforms, are flexible and if they can monitor, their platforms, and monitor their hardware remotely. So um. You know mapping just mapping sort of the use cases, on the spectrum. Some things are at quite an early stage when it comes to sort of autonomous, driving, and, you know traffic, managing, traffic, seeing, sort of automatically. And some of these more custom engagement. Very arvr, heavy applications. But there are, use cases and applications, which are more advanced, for example asset tracking. That has been using some level of edge compute, in the, in the vehicle for example for a long time.

Um, Or, car gaming we're starting to see, car gaming. Um, being deployed in regional data centers which are a form of edge you could argue. Um, so you know, there are things are moving quite quickly. Applications, are moving sometimes other technologies, need to come together. And sort of advance whether it's in the iot space or arvr. To sort of push things along. And i guess the last thing i wanted to touch on is just. The. Edge compute. Challenges, and this range from, you know some of these are some of the barriers to overcome this ranges from sort of, organizational. Ones and non-technology. Challenges, where, you have conflicts, between 180. For example. And that's something that's prevalent. As say manufacturers. Move to industry 4.0. Um. You know and in general. In other you know in other sectors, say oil and gas as well 1890. Have been quite distinct, now with edge they're coming closer together. And there's you know there's a challenge of who's the decision maker there is it someone who heads up it, someone heads up ot is there a need to evolve that, that. Organizational, structure. Second thing is about culture, and education, as with any you know change. This takes time it takes time for. Um. Enterprises, to implement these new applications. So, you know for example predictive maintenance. That, um. That fundamentally, changes how maintenance, is conducted. Um. And, that takes time to train people about it takes time to make them comfortable, about it and then you know to allow them to give them the tools. To, to use that technology. Third thing is data compliance, regulation, i think i mentioned this at the start but that's becoming, increasing. Uh increasingly, a challenge. And the challenge, is um not just being aware of, different regulations, so say for a multinational. They need to be aware of different regulations different jurisdictions. But they also need to be able to manage their data seamlessly, and today it's quite manual in the sense that, you they sort of either store data in a in a data center. In india for example, or in china they have to manually, move that data or in, the us. Um or in europe. Um. What you know ideally, what should happen is that. Things are, automated, in a way so you can, almost label data and that data, moves. Um, in the right edge of the in the right cloud depending on the needs. Of the enterprise, and also, depending on who the user is and where that you know which jurisdiction, is relevant to that user. And then the last ones are a bit more sort of technology. Focus, where um the first thing is that. You know because, of this move. From, you know a lot of it's coming into ot and that's a big part of edge moving that that, using the i.t like infrastructure, for ot. In an on-premise, environment, in some kind of edge-like. Environment. Um, but the challenge is that. It's not drag and drop and that there do need to be some changes, so this could be about things like, making, equipment, more ruggedized, the server's more ruggedized, so that they can withstand, the environment, in, harsh conditions. Um, you know for example in a remote, um. Oil rig. Or it can also be about, making it suitable. For, the customers, needs, um, in almost the opposite way where again like taking retail, or, i don't know for example, um. In a, you know in a kind of a. More hospitality. Environment. Or in a, stadium, or you know whatever it might be, you don't really want, huge, deployments, of edge compute physical deployments, you need things that are small. Um that can you know slot in easily. And don't take up too much space. And then. Second last thing is about making the business case, um. Enterprises, do need to have a think about you know what's going to be cloud what's going to be edge there is a cost associated, with both. Um. It can be difficult, and you know a lot of these industries, are already, cash trapped they're already, in a hugely competitive, environment, so it's very critical. That there is a clear business case for changing, how their infrastructure. Um evolves. And then the last thing is. The partner ecosystem.

And Again you know edge is still relatively, new and it's not so clear. Who, are the leaders in edge or you know which who to partner with. And it's quite fragmented. Because you need you know you almost need different parts of the ecosystem, to bring it all together. And this diagram, here, sort of shows that where this is you know this is kind of essentially the valley chain of edge where you know you have, the. The facility, the premises. The network, part, or connectivity. Hardware. Software infrastructure, etc. All these need to come together and there are specialists. In each domain so it can be quite confusing for an enterprise. But it is quite, important, to pick the right partners to to build your edge compute. So what to do well first thing i guess you know is to build that business case to find the key drivers. The kpis. Figure out which applications, need edge. Um. Understand this difference between itunes, o2 and you know how to bring those together. Um whether there's an also an organizational, structure that needs to take place, um that you know that's almost its own business case in its own right. Second thing is defining some guiding. Principles. Um. To make sure that the plan is future proof. Um. So what i mean by that is that you know. What what, what edge computing, is allowing, enterprises, to do, is to learn from the mistakes of the past where, traditionally, it's all been about proprietary. Systems, and that has got. Indus. Enterprises, in certain industries. Um in quite a difficult position, as they, have had external, pressure to evolve, and, they are being disrupted. But they're unable to change because their their physical. Or their itnot. Infrastructure. Is not flexible. Um so it's important to make sure that even though. Maybe the applications, for edge are not clear yet and it's you know there is an unknown, for the future, but to actually, put that foundation, in place that will be future proof and will allow the, um, the enterprise, to be flexible, and to change things, going forward. And then the last thing is you know about selecting the right partner, so all these different there's lots of different technologies, coming together here. Not just in edge but there's sort of 5g coming in the horizon, iot is a big part ai and ml, and arvr. And all these things, and so again, you know, you need to make sure you have the right partners to understand these technologies, and not and are able to. Um augment them into their solutions, if not today then in the future. Um, and the other thing is that you know, it's great to have partners who can bring others together. So. A lot of. You know a lot of companies, now some companies, are really taking a role in, bringing, the ecosystem, together and building that ecosystem. Um and that's across software, and hardware, and services, and that sort of thing and that's kind of critical for this, too. So that's it for today thanks everyone, for. Listening, i hope you, found it useful. I'm happy to, take any questions offline. Thank you very. Much. You.

2020-08-21 02:04

Show Video

Other news