 Hello and welcome to our session. I could be calm in clown native calm North American 2023 I'm Tina Joe and advocate for infrastructure ecosystem advancement at arm and I also have the owner of serving as the board chair of LF edge in collaboration with my colleague ending from Google today we will take you through a comprehensive guide on leveraging Kubernetes for edge computing this guide is crafted to help you to understand how Kubernetes can be optimized for the edge to spearhead innovation in clown native ecosystem the best of practice that are essential for creating a resilient and scalable edge infrastructure this covered convergence of our collective efforts and promising the future they hold for technology at the edge we are excited to share our insight and look forward to a productive session focused on the robust possibility that Kubernetes brings to the edge computing welcome to our session on Kubernetes at the edge part of a could become in clown native calm North American 2023 today we will navigate through various aspects of edge computing and its growing intercession with Kubernetes our agenda includes understanding edge computing we will begin with demystifying edge computing defining what it is and discussing its importance and unique characteristics why Kubernetes in edge computing next we delve into the reasons Kubernetes is becoming a popular choice for managing for managing containerized applications at the edge highlighting its benefits in such environments Kubernetes projects on the edge we will explore several Kubernetes projects that are tailored for edge computing providing insights into how they operate their use cases Kubernetes on the edge in production this session will showcase real world examples of Kubernetes deployments at the edge emphasizing the practical consideration and best of practice LF edge catalog we'll take a look at the LF edge landscapes focusing on the catalog of projects and resource available for those looking to implement Kubernetes at the edge considerations for the future to conclude we will discuss future change potential development in edge computing and how Kubernetes expected to evolve with this space let's embark on this journey to edge computing with Kubernetes understanding its framework applications and exciting potential it holds for the future as we delve into the concept of edge computing let's visualize its ecosystem edge computing brings computation and data storage closer to the sources of data this proximity to data at its source can yield numerous benefits including significant reduction in latency and bandwidth use here is a breakdown of edge computing layers illustrated in the diagram near side edge we call your neck this is the closest edge layer to the end user located at home enterprise or in node b levels at this layer we are operating on the scale of millions of devices with latency as low as five milliseconds suitable for immediate response applications like v2x arvr and ai inference city edge as we move further from the near side edge to the city edge the scope narrows to hundreds or thousands of nodes serving applications like content delivery network cdn and more intensive arvr tasks regional edge expending our reach the regional edge integrates cloud services and can serve tens to hundreds of applications dealing with tasks that can tolerate slightly higher latency such as ai training big data analytics and video transcoding cloud at the broadest layer the cloud encompasses services that can handle larger latency over 20 milliseconds suitable for less time sensitive tasks this multi-tier architecture allows for a wide variety of applications scaling from millions of devices at the edge to tens of nodes in the cloud optimizing the network to serve specific applications needs based on latency data volume and processing power required edge computing represents a paradigm shift in how we process data by moving computation closer to where data is generated we can unlock several key benefits reduced latency edge computing allows for faster response time by processing data locally rather than sending it back and forth to a distant cloud this means actions can be taken almost instantaneously after the data is produced which is critical for applications requiring real-time decision-making such as autonomous vehicles and industry automation bandwidth efficiency with local processing we send significant less data across the network this not only speeds up the data analysis but also saves bandwidth for other uses and reduces network congestion leading to a smoother more efficient digital experience enhance the privacy and security when data is processed locally and not transmitted over a network there's a less opportunity for interception or ifs dropping this inherent reduction in data travel boasters security measurements measures and enhanced user privacy by keeping sensitive information within a confined local network understanding these advantages we can appreciate how edge computing is not just an evolution in computing architecture but a necessary step towards a more responsive efficient and secure technological ecosystem in the expanding realm of edge computing kubernetes immerse as a cornerstone technology here is why kubernetes is essential scalability as we push the boundaries of edge computing we are deploying not just a handful but thousands of edge devices kubernetes excels in handling this scale allowing for the efficient management of extensive networks of devices without compromising performance flexibility the landscape of edge computing is incredibly diverse kubernetes brings the adaptability needed with its seamless deployment and management capabilities whether you are dealing with various workloads or different types of hardware kubernetes can make it easier to deploy applications consistently and manage them efficiently resilience reliability is non-negotiable in edge computing kubernetes introduce self-healing mechanisms such as auto replacement or failed nodes auto restart of non-responsive programs and even automated scaling to ensure consistent performance this resilience is vital in maintaining the robustness required at the edge of network where physical access can be very challenging kubernetes doesn't just help to manage the complexity of edge computing it's enabling organizations to harness its full potential by offering a platform that's as dynamic and distributed as the edge environment itself adopting kubernetes for edge computing comes with a set of our best practices to maximize efficiency and security let's go over these key practices modular design simplify your applications by breaking them into smaller and dependent modules this module modular approach is ideal for deployment at the edge making your applications easier to manage and update state management and for stateless applications or synchronize stay carefully to prevent data inconsistency proper state management is crucial for ensuring high availability and reliability network optimization keep data transfer to the bare minimum without sacrificing necessary communication efficient data transfer is vital for maintaining speed and reducing latency in edge computing environment security prioritize security by securing device identity and use end-to-end encryption regular patch management is also essential to protect against against vulnerabilities resource management efficiently managed resources by optimizing cpu memory and storage usage kubernetes can help automate this optimization ensuring that your applications run smoothly logging and monitoring implement centralize the logging and real-time monitoring these practices are key for quickly identifying and resolving issues maintaining the overall health of your edge computing infrastructure by following this best practice you can ensure that your kubernetes deployment is as robust secure and efficient as possible in an edge computing context thank you tina so i'm continuing talking about this uh edge developers are using kubernetes so this is the story we did on 2021 by since f so it shows more than two-third developer in edge area already using kubernetes so uh we are doing a new one in 2023 during the co-ed but you definitely can see the trend why edge developers are using kubernetes because uh the things tina just conclude i won't elaborate but you can see the trend i think people can decide by their foods so uh here are three popular kubernetes projects in cncf there's a micro kubernetes is developed by the canonical is a fully component kubernetes with very small footprint and also k3s is developed by the renter is an official kubernetes sandbox project is developed by renter and also the kubernetes edge this uh we won't call is a kubernetes distribution because it's break down the kubernetes by cloud side and edge side do a few modifications so it's extend kubernetes in the edge computing i will elaborate at least one more in the following slides and also there's many more and also for example the open yard is developed by alibaba is very popular too so we did have a panel discussion in cncf the cloud native count 2021 however it's what a virtual uh hybrid events in la so if you are interested you can come to watch the video we invited people from renter, salesforce, alibaba cloud and all these people we discussed about compare pro and cons on the user cases for kubernetes projects in the edge computing if i interest you can look at uh watch the recording from 21 2021 cloud native count kubernetes so we plan to do another one next year's european kubernetes count to refresh the knowledge to see what's going on and what's the update so i i'm going to elaborate a little bit on the kubernetes why we use this so we use the extended kubernetes because they give us a unified deployment apis so basically before that if you want to deployment on the edge so you need to do your own ota update however with kubernetes so think about that you have you are a regular cloud application developers so what you do is just do the kubernetes control deploy or kubernetes control apply it will do that for you that's the separate from the kubernetes edge in order to solve the edge cloud communication problems we set up a channel so we replace the kubernetes apis uh apis server with a cloud core basically it at the proxy talk to the edge so instead of a simplify use the kubernetes we have a kubernetes the edge core is extended from kubernetes so they take over the communication between cloud and edge node so that solve the network problem so because it's not a kubernetes session i won't go deeper enough but with this application so you can get a benefit from kubernetes the uh unified kubernetes deployment apis uh auto scaling and also resiliency so here are some already in production examples so first is on the left basically from the cloud you talk to a highway tow booth so each tow booth you deploy as an edge node so you can see that a heterogeneous deployment what you could be the node could be running on x86 servers or it could run on arm servers because with kubernetes you can internalize your application on the edge so you compile different cores then you deploy on different nodes however the apis are the same you can accumulate data back to your cloud and do the analysis on the right is a bridge monitoring you can see the real picture how you place your edge node your sensors your devices how you communicate and use a 5g or 4g technology to communicate with your cloud you do the analysis and monitoring so here is a little bit elaborate how the actual deployment is so in order to deploy on the edge node it's impossible to go in the 100 kilometers road to update individually so and also you don't need to worry about your developer your own OTA update programs what do you do is just kubernetes apply or kubernetes update so your kubernetes kubernetes control apply a new version then your edge node application will rolling out using the same kubernetes deployment mechanism another interesting thing is a acreno acreno is an alpha edge project so tina is alpha edge tsc board member board chair i'm the tsc acreno tsc chair so we have a bunch of blueprint basically the blueprint is a combination of binary deployment to showcase how your edge computing should do and you can download the binary to verify yourself and that's all open infrastructure the application includes 5g ai and an interesting thing is that we use this one is at the edge ai even though uh this is the architecture it showcase you how you put your inference app on your device so because it's more close to your to you and without trying for all your data back to the cloud for save your bandwidth and also uh low latency and privacy so you don't need to upload your sensitive data back to the cloud it depends on where your edge located if you have a large at a city edge you can deploy large models for example lm a large language model you can deploy there however if you want to deploy on your home or even on your cell phone you probably have to have a small tailored model to deploy there we showcase this one we are going to upload this uh slides you can download from the links then you can try yourself to see how you can deploy a new application inference application on your mobile device to do the inference and also how you can update your model through a simple api without complicated your life yeah just to add a comment to the uh ai edge there so we do have a lf uh edge ai edge proposal and uh with the hyperscalers join us bring in their own large language model our first showcase will be using the open ai whisper model for the voice recognition and uh we probably do some fine tune for the verticals and we'll showcase with hugging face and so welcome to join we have uh apm california time every monday we have meeting you can find the wiki page on the lf edge today uh we are delving into the lf edge catalog and advance the solution designed for managing linux foundation edge applications with ease and efficiency comprehensive uh uh repository the lf edge catalog stands as a comprehensive uh repository centralized the resources and simplifying deployment uh processes for users it emphasizes on harnessing edge computing for potential without the adding complexity catalog features it's just like you're walking and for a store is a can i try on for two days here you go you got a car you try on every single aspects of the car for two days the catalog exhibits the smarter application stack reflecting its adaptability and breadth it demonstrates how the catalog supports diverse iot smart cities and emerging technologies applications by making dunking tasks manageable catalog impact this tour is instrumental in driving the evolution of edge computing highlighting essential areas such as iot and smart cities it's designed to transform complex challenges into streamlined processes and fastest innovation in the edge computing domain target audience is the lf edge catalog is tailored for developers and it professionals it acts as a bridge for those inch guilt by edge computing it providing a platform to explore an experiment call to action we encourage you to utilize the lf edge catalog to facilitate more efficient operations and departments it's an open innovation to partake in shaping the future of edge computing with the lf edge catalog as your guidebook yeah i hope you enjoy driving the car for two days but leveraging the lf edge catalog you position yourself at the full front of edge computing innovation prepare to tackle the challenges and opportunities that come with this rapidly evolving technology landscape as we look forward the horizon of kubernetes edge computing there are critical considerations that will shape our journey ahead interability the key to a seamless connected future lies in interability with an ever-expanding universe of devices ensuring they can communicate and work together is non-negotiable it's about creating a cohesive ecosystem where compatibility is the cornerstone sustainability our technology must not only be smart but also sustainable we are steering towards incorporating power efficient devices and harnessing renewable energy sources like solar panel solar power for edge devices the aim is to minimize the carbon footprint of our technological advancement regulations i see somebody from government is sitting hi so staying updated with the evolving landscape of data privacy laws is um imperative as regulations tighten globally our systems and operations must adapt to maintain the compliance and protect user privacy these three pillars interability sustainability and regulations will guide us as we navigate the challenges and opportunities in kubernetes edge computing by adjusting these areas with foresight and innovation we set the stage for a resilient and responsible tech ecosystem as we cut imitate our discussion on kubernetes and edge computing it's clear that these technologies are not just parallel checks but are converging to propel cloud native innovation forward unify the progress kubernetes serves as the orchestration backbone while edge computing edge computing extends the reach of our cloud capability together they are a synergistic force ushering in a few error of disability computing that is more accessible efficient and scalable best practice integration embracing best practices is more than a recommendation it's a necessity for creating an edge infrastructure that is resilient and agile by committing to standards and continuous improvement we can navigate the complexities of modern computing environments confidentially or confidentially i mean in summary the intersection where kubernetes meets edge computing is fertile ground for technological growth by forcing collaboration prioritizing innovation and adhering the best practice we position ourselves at the forefront of transformative wave in the industry here is to a future where the potential of cloud native technologies is fully realized in a world that's always on and always connected thank you all for your participation and attention today and it's been a pleasure discussing kubernetes for edge computing and its impact on building resilient and scalable applications we hope you found this session insightful and that it sparks further innovation in your projects we do greatly appreciate your feedback to continue improving and making these sessions more valuable for you please take a moment to scan the qr code display on the screen and share your source on this session your input is crucial for us to learn and grow and we look forward to hearing from you once again thank you for joining us at kubernetes and cloud native calm north america 2023 i see somebody going to the mic no yeah we have five minutes left if you have any questions feel free to go to the mic thank you