hi there my name is dave chacochus with lumen and in this most distant most abnormal of years uh it's still great to be with you here at re invent at 2020. uh we're excited to talk to you today about a great important topic that's starting to become more and more relevant in the enterprise which is distributed applications and edge computing and some of the patterns we're seeing play out on our network and within our work with amazon web services and we sort of boil down a few key issues around sprawl security and strategy that are occurring at the edge of the network and are occurring in some of the workloads that customers are using in a number of different industry scenarios and so we wanted to talk a little bit about what we're starting to learn and how we're working with aws to address some of those trends out at the edge well before we do i wanted to make sure that you understood who lumen was if you weren't aware of lumen and the company um that uh that we are we're a large global communications company in fact we have one of the largest fiber networks in the world over 450 000 route miles of fiber that can deliver very low latency connectivity to a broad range of enterprise locations multi-tenant data centers cloud-on-ramps such as those with aws and their direct connect program how our network is also very deeply peered we're connected to thousands of autonomous systems which means that traffic on our network can reach those eyeball networks and be efficiently routed to a lot of the different uh the end customers that are going to be working with a lot of these distributed applications but in addition to that public networking peering in public networking capillarity we have a a very deep and very broad private networking capability that allows us to provide optimized path optimized routing dynamic connections to any one of the vertices on our network through software-defined networking and that includes uh amazon web services in addition inside of our network we're starting to become starting to become more and more prominent to have edge computing resources inside anywhere one of our network nodes intersects we need those edge computing nodes in order to deliver value-added services but we're finding that customers also look to our network to have computing resources available to them for some of these distributed applications and some of the workloads that we're going to talk about here in a minute so we have an edge computing capability that can deliver up computing resources some of them are you know from from lumen as a first party service but we also have the ability to go and co-locate an aws outpost inside of any of the nodes of our network and we'll talk a little bit about how some of those use cases can start to play out so what lumen and aws really have in common is that we both power the fourth industrial revolution right those whether you go back into uh the early uh time frame of of steam power in the first industrial revolution that that enabled uh to the to the innovations around electricity and electric electrical distribution systems to uh the age that we're largely now in with this sort of third wave of computing around technology uh that is that has really enabled companies like aws and cloud computing and data center virtualization really the merchant silicon era of of our industrial revolutions but what we're really starting to see now and this idea of a fourth industrial revolution is a revolution around data access to data being able to acquire analyze and take action against data very rapidly in combination with and in complementary to the tremendous computing capabilities that exist inside of the cloud and the centralized data center footprint so lumen and aws are close partners we're an advanced consulting partner inside of the aws partner network we can deliver managed services and we can deliver migration solutions we can work with them in some of our public sector contracts we have a deep direct connect capabilities around aws and we work with a wide variety of enterprise workloads running up inside the aws platform and really the work that we do throughout the network all the services we can add to the network and then all the ways that we can route to aws can really introduce some really valuable powerful outcomes for our customers together with us in the aws cloud platform but what we wanted to talk a little bit more about today was this idea of the edge computing and how so much of the fourth industrial revolution all these industrialized use cases and smart manufacturing and smart facilities and smart cities are starting to treat planet earth more like a bit of a motherboard where the network connects all of these locations together and so many more of the applications that we see running in the fourth industrial revolution are those that have to cover broad geographic distance and need to put workloads in closer proximity to the digital interactions they're aiming to control and so that's really why lumen exists right we we are one of the we believe that one of the the powering engines of the fourth industrial revolution because of how much uh geography to geography communication is going to be required so really excited to think about all the opportunities that this fourth industrial revolution can lead us to uh whether it's with new industry innovations asset intensive industries that are producing tremendous amounts of data as well as a lot of the human immersive technologies and the kind of digital interactions that human beings can embark upon with all the different types of business models and other people and other things that people can interact with tremendous amount of opportunity but there's also a large number of considerations that need to be taken into account when you're thinking about your enterprise workloads and how they're going to be distributed throughout the network how they're going to be distributed throughout the physical footprint of your enterprise and so we alliteratively boil those down to sprawl security and strategy and we'll sort of touch on each one of these three talk a little bit about how aws and some of their edge computing products in combination with what lumen can do from the managed services and a network standpoint can really combine together to address a lot of these concerns together so from a sprawl standpoint it really boils down to the locations you're thinking about uh the ability for your innovation pipelines to reach all those locations and then how you're thinking about investments and being able to capture and reuse the hardware you deploy over time so when it comes down to edge computing and when the edge computing workloads are starting to spring up and the ability to go and acquire analyze and act upon data how closer to the edge of the network or closer to your production facilities what you're tending to run into is a lot of unique locations right you're certainly way beyond the cloud at that point right in a lot of times you're working in somewhat unusual environments environments that didn't usually think of themselves as data centers they're going to have limited i.t staff they're going to have a varying degree of how dynamic degree of dynamism that's occurring across those environments where you're going to be having production workflows lots of physical activity uh potentially a constrained amount of space or cooling or power so some of the facilities uh and then the location diversity is an obvious and primary challenge of edge computing so thinking about how to take advantage of the scarce real estate you have we're also going to find is that for edge computing to be powerful edge computing needs to be able to analyze and take action upon the data signals that all of the different things and people and business models are capturing out at that edge location and so what that has historically led to especially in some of the early days of these iot use cases or internet of things uh what you have tend to find is a lot of one-off design concepts a lot of technologies that are in the iot arena that are really well geared for sort of walled garden technologies uh the ability for a bunch of devices talking on a flat network to be able to control um a range of things via policy and what that does what that tends to not lend itself well to is that kind of agile software software driven pipeline driven uh innovation pipeline that is the hallmark of cloud innovation and cloud computing and so really thinking about how you create software how you analyze act uh and acquire more data to be able to improve your business models is a key factor to consider does your software pipeline and how can your software pipeline reach all these locations because if you're distributing your applications if your applications are increasing in their diversity in their footprint making sure that your software pipeline can consistently reach all those locations is a critical concern and then all of the different hardware form factors that are going to be going on out there at the edge companies are really struggling with and thinking about how can i make sure that hardware that i put out there is going to be something that meets the requirements and the needs of what i'm trying to build out of the edge of my network or inside of a physical premise facility but how can i make sure that i can reuse that hardware uh you know the aforementioned iot use cases there's a lot of specialized hardware that can be used to go and do signal controlling um and do programmable logic that can be fed into a lot of those internet of things devices but that hardware tends to become very specialized right and how can companies are starting to think about and ask how can they more efficiently get access to that hardware so that it can be used for a range of different purposes that will evolve over time can they is that something that they can load into a small form factor piece of telecom premise equipment is that something that they can host up inside the network in order to be able to achieve that hardware as a service in order to think about their edge computing outcomes and so all of these sprawl challenges are things that are on the mind of our customers but sprawl fundamentally comes down to priorities right is solving the problems of edge sprawl where does it sit on your priority scheme and you know we talk to our customers all the time we've been talking to them all year we tend to talk to them in the context of some of these edge computing edge sprawl types topics and i wanted to share with you a few of the questions that we commonly ask throughout our enterprise engagement with our customers around how they're thinking about distributed distributed applications and the sprawl of those applications across their estate and so this is one question we tend to ask which is uh how important is overcoming sprawl to you how important is driving digital workloads closer to the interactions that they control you know and runs the gamut from answers from the gamma from very important uh we absolutely know that not everything can be centralized or it can range down to you know we're not even really thinking about that we're still at the very early days of thinking about how edge computing might come into play and so what we tend to find out and this is you know a small sample set it's a sample set with our with our enterprise customers that we've engaged with at a range of industry events not as big as aws re invent but throughout the course of 2020 and our customers tell us that by and large no one's uh considering that edge computing is unimportant but that well over half of customers are already thinking yeah there's probably an opportunity here for us to take advantage of the the data we control the assets we control the digital interactions that drive value for our business we feel like having a distributed footprint will put us in a better position and it's something that we absolutely feel like we need to contend with so 50 or more think it's important or very important so that's the sprawl challenge the second main challenge that we find our customers working with is the security implications of a larger distributed application base and those boil down to network patterns the larger physical footprint of technology and then the broader data footprint of their technology so on a network pattern basis you can think of customers thinking about distributed applications edge data centers the workloads running inside of a particular enterprise premise as you know being doing that for a reason and the reason is that there's a workload that needs to run and it needs to get access to the network where the digital interaction is occurring and so in many ways that can lead to making sure that you're running a workload on a well-peered network with access to lots of eyeballs um sometimes it'll mean making sure that you're running that workload inside of a smart manufacturing facility on a uh on a wireless network with very clear channels the workload network itself is a whole factor of workload design making sure you have clear and consistent and secure and reliable communication between all of the digital things that are digitally interacting on that particular network whether they're people whether they're things or whether they're business models but equally important is making sure that you can get the right management access to that environment that you can push new code to that as part of your development pipeline that you can pull uh telemetry out of that environment and feed it into your operational systems so not only the workload itself but management access to that environment is something that turns into its own design exercise and then all roads still tend to lead to something centralized whether it's an enterprise data center or a public cloud region and so making sure that you have good connectivity back to the cloud on a secure reliable channel is a whole range of network patterns that need to fit into the company's existing security and risk management strategy what another element of that secure another security dimension is the idea around physical footprint and all the different technology that's going to be part of a lot of these workloads that are creating all this very important and potentially valuable data are all now subject to tampering or thefts or people physically manipulating those environments it's a broader footprint of things that could potentially become infected it's a broader range of vulnerabilities that come from exposed usb ports and and potential manipulation of the device technology that's out there in the field and then the final piece and probably the most powerful piece is the broader data footprint that edge computing exposes enterprises to a lot of the data that comes from a lot of these highly distributed things and highly distributed digital interactions can oftentimes become intensely personal you know voice recordings and video images and characteristics that especially when correlated together can really start to become uh a high degree of of exposure in terms of privacy and due care that a lot of enterprises may not have otherwise thought of until they start to see the data getting correlated together so being able to understand strategies for processing data in a stream and not storing it looking for digital events or signifiers of a particular point of value without necessarily having to compose all of the the data elements and and maintain them and care for them over time having specific policies around different devices and how they can capture collect and store data and be able to manage those policies in a broad distributed location is a big part of what enterprises think about when they think about the challenges of security and that broader data footprint and so again we'll talk a little bit about here about some of the feedback we've got from our customers as we've talked to them about edge security and we asked them you know of those three dimensions you know which ones concern you the most uh you know is it the making sure you have your network securely connected to the right framework is it making sure that your devices and your device footprint and all those smarter devices in the field are well protected or is it really about the data liability and collecting and protecting data from so many more things introduces its own dimension of risk and as you might expect when we talk to our customers and have surveyed them throughout 2020 the responses to that survey really break down equally right it's about about a third or third a third although if you had to index in one way you'd see that more of a few more customers and that 36 percent uh out of the sample set have given us feedback that is really getting the network design right that is probably the thing that keeps them that has the most concerned and making sure that they have good secure network connectivity to all those devices although the other issues in the other dimensions are are certainly very important so the final thing that we talk to our customers about when it comes to distributed applications and some of the architectures that they can implement in order to be able to manage those dynamics is really having a good consistent strategy for how you're thinking about distributed computing and distributed applications that stem from the global cloud core all the way out to the distributed physical edge and there's really four common functions uh you know tom bittman over at gartner likes to talk about and use phrases like the law of physics the law of economics the law of the land and murphy's law those four fundamental forces are tend to be what drive a workflow closer to the edge right you need deterministic latency which is actually uh can be potentially a very powerful driver but you know the latency latency on a fiber network and the blink of an eye is still you know a few hundred milliseconds and that's that's a pretty low latency uh you know duration for the vast majority of workloads it's when those workloads tend to compound and it's when thousands and thousands and millions of those transactions are all compounding with each other that obviously uh the certain deterministic latency budgets are necessary but more often than not it's not latency it's actually a combination of some of these other factors that are driving workloads out closer to the edge it's the economics of just all the the data creation and the bandwidth management moving all that data around sending it all to a centralized data source or a cloud platform can actually blow the budget and overcome the value of what that data flow is even producing there's also physical there's physical locations and statutory requirements of what your local government will allow you to collect with some physical specifics around where you're going to be tracking that digital interaction that can come into play so from a compliance and a legality standpoint and then there's just the the overall uh autonomy of a particular facility some facilities that are introducing new innovative ways of managing data and introducing business logic are production facilities they absolutely cannot be reliant on a network connection or a cloud connection no matter how secure how redundant they don't want to even have exposure to that potential loss of control and so they want to put the workload on site because they need to be prepared for anything well those four forces tend to get focused on these idea of digital interactions and i've said this a few times now here in our discussion are around the digital interactions between people between things and between business models and all the different 12 or so ways that you can combine those three variables um and think about the ways that those uh different types of digital interactions can add value for the enterprise so it's those four forcing functions between those three types of digital interactions that are fundamentally all producing data that needs to be acquired analyzed and acted upon and how fast you spin this cycle is fundamentally a measure of how much of a digital business you are um you know again getting back to the way we talk to our customers and the the survey question that we like to ask is and this is sort of a basic one what is your level of innovation velocity with these types that digital cycle of acquiring analyzing and taking action upon your data and that'll range from companies that really consider themselves expert in terms of digital velocity multiple deployments a month across all the data they have control over and that ranges all the way down to a company with a very limited digital profile and that barely even thinks of themselves as a digital enterprise and the results we get from our customers is that the majority of them are still very early on the journey um you know they'll give themselves an answer right at the middle of the pack if not a shade lower uh in terms of a moderate uh satisfaction with their level of digital velocity and what that certainly suggests to us because we collect this data as we're talking to customers about digital workloads occurring at the edge of the network that are running closer to digital interactions a lot of these are very future forward fourth industrial revolution use cases which we'll talk about some examples of here in a second but when we talk to our customers about that and they still reflect upon the fact that they're still generally very moderate in terms of how much of a digital enterprise they are it tells us two things it does is not only are they interested in the future they still know that they have a ways to go in the present and so really working with cloud partners and cloud platforms like amazon web services and the kinds of partnerships that these customers need in order to take their business into into overdrive is certainly top of mind right really turbo charging that cycle you can kind of think of the cycle that spins around the enterprise data of acquire analyze and act can really be thought of as being sort of supercharged uh by edge computing because you're acquiring data from many more sources you're analyzing those patterns in real time and you're taking action on business logic closer to that digital interaction and we absolutely see examples of our customers already adopting these patterns especially when there's an asset intensive or an interaction intensive industry so we want to give you some examples of where we are starting to see our customers um start their engines and we're going to give a couple of examples here about how amazon web services can really fit in uh to some of those profiles uh where we have one of our customers that is in the logistics business they have thousands of physical sites across the country and they have a forcing function that they're trying to achieve that is focused on the digital interaction between things and business models where they have a lot of material flowing through their facilities they want to be able to not only do some automated routing of unmanned vehicles inside those facilities and know exactly where and the best way to position and logistically flow material through those facilities they also want to be capturing high resolution imagery and being able to do analysis for security um and loss prevention uh within those facilities at very high resolutions uh they can get better insurance rates if they have good uh you know telemetry based analysis of devices and when uh when when material what shape it was in as it moves through their facilities so there's a lot of business logic that can go take action against a lot of the enterprise data that's falling through this facility but it's really forced by the idea of a combination of low latency milliseconds and high bandwidth magnitude that is the forcing function for them to consider actually building edge computing infrastructure inside of or in very close proximity to those thousands of locations inside of their nationwide footprint we're seeing other customers and this is an example of one we're working with that is out inside of the geotechnical mine uh where they have a lot of interaction between physical things the earth that they're mining against and then all of the heavy machinery that is moving through that site and those sites tend to be any any of you who have ever visited a mine um know that the site uh for some of these geotechnical environments are larger than some cities huge expansive footprints across which devices and material and in some cases unmanned material is moving through that overall mine site and so the ability for them to be able to control and run business logic in close proximity to and break out some business logic that is going to run in very close proximity to the low latency network necessary in order to run that facility has led us to start looking at private cellular and private lte type mobile networks that you can run within a breakout facility in the nearby network where you can control all of the access points without having to run wi-fi networks and do all the handoffs necessary between wi-fi access points which are in a constantly shifting physical landscape and sometimes in a explosive landscape in and around that large physical footprint so communications in order to help autonomous vehicles is an example of some of the use cases that we're talking to customers about obviously uh the digital interaction between people right people with people in different types of collaborative business models or we're also starting to see a lot of interest and interaction around is in the gaming industry people interacting with other people in business models that are based in fairness and competition in some kind of a digital online form factor in an online game really your driver there is both milliseconds and then being able to drive business logic between potential players so that when a game is blowing up in a particular geographic area being able to get very specific infrastructure into that geography so that you can create a fair game between all the players in that arena is is introducing a lot of opportunities for gaming companies to explore edge computing and thinking about moving those workloads closer to the digital interaction between their players and then the final one we'll dive into here and this is really an example we think of where something like aws outpost comes in and can be potentially very strong is in and around some of the smart manufacturing and supply chain use cases we're seeing on our network we recently worked with a large national restaurant chain who was doing some supply chain integration into their agricultural source of supply and they built a very sophisticated automated smart manufacturing system to be able to process that agricultural input into their facility and they're really focused on the business model outcomes of better efficiency lower waste uh and then capturing uh the maximum percentage of all the inputs and turning them into something valuable instead of capturing the raw inputs sending them around the country and then letting a lot of it go to waste right so there's dynamics of uh the ability for this to occur at very low latency with all of the machinery controlling all of the device flows through the facility as well as that sort of murphy's law autonomy dynamic of this facility can't go down we can't have this thing offline it cost us potentially millions in terms of critical inputs into our business and so we have to run this smart manufacturing facility with some kind of a local control and this is a great example of where something like aws outpost working with a network provider can clearly come into play uh where you know where we have in the architecture that we've looked at with this account that would be a great fit for something like an aws outpost running in the computer room on premise talking to all of the devices over that local workload network via a lumen managed wi-fi link with the managed services being able to go get that environment set up get it wired into that local network and then using the broader lumen backbone to be able to create the kind of management networks necessary to get back to the aws region where some of those longer term data analyses of all of those business outcomes and critical metrics can be measured with great things like aws sagemaker and redshift and the the data analytics modules that exist inside of that centralized model and we can go and run that outpost both at the customer facility or it's also something that we could run from inside the network one hop away in order for that local environment if they didn't have the computer room on premise they could run it from within the lumen network to achieve largely the same outcome at a very low latency reaction time so as we think about uh all of the companies that are on this journey first off you know we're all here at aws re invent 2020 we've come a long way from uh you know the early days of aws re invent when we're still talking about what is cloud um we're definitely at a point where the enterprise is thinking about not only continuing their maturation of adopting digital technologies and agile innovation factors throughout their business but they're also taking a look at this is going to start to get a whole lot more distributed depending on how asset intensive or interaction intensive their value creation equation is and so getting started with partners and help companies that can help you acquire data from all the places that their enterprise reaches analyze it better act upon it faster while keeping it all secure is something that a lot of our customers are thinking about lumen can go and do implement solutions like the one we just talked about there with aws outpost in any one of the network vertices you know they're listed here these are a list of all the sites where lumen has onnet where we can do an edge computing solution we have some areas where we're pre-staging compute capacity where we can go and spin that compute capacity up very fast for some small workloads and then expand out and work with companies like aws in any one of those locations to deliver workloads in closer proximity to digital interactions for all of these fourth industrial revolution use cases and so the final point would wrap up with here is that to go back to our survey topic is that enterprise conversation we're having around what kinds of strategic partners are most important to you what are the strategic partnerships you are thinking most about as you're starting this journey to think about how to move your workloads up closer to the edge you know are you thinking of the the cloud service providers like aws are you thinking about your mobile enterprise mobile provider are you thinking about somebody that does run a backbone and can deliver a network solution that can optimize all the different routes through the network and all the different locations that you can already reach are you thinking of the systems integrators are you thinking the colo providers we ask them these questions and we let them choose you know whichever they think are most strategic and important and the type of partners that tend to shine through as you probably imagine are the hyperscale clouds are the companies that are helping companies think digitally and move faster and really take innovative steps to take software and rapidly improve it in a cloud automation style but what also shines through the other provider that starts to bubble to the top are the network solution providers companies that can go and solve all of those different network flows and security challenges while maintaining some control and giving them optionality about how to be most efficient with their sprawl so with that uh we'll wrap up there and hopefully that gives you some ideas around the kinds of companies that would be helpful to your business as you're starting to think about distributed applications and taking your cloud presence and be able to move it out closer to the edge of the network moving workloads closer to digital interactions for these next generation use cases that we think are really the future of what the fourth industrial revolution is all about thank you very much for your time hope you enjoy the rest of re invent
2021-02-13