HackMD
Akri KubeCon NA 2024 Co-Located Events CFP
https://events.linuxfoundation.org/kubecon-cloudnativecon-north-america/co-located-events/cfp/
Dates
Salt Lake City, Utah: November 12, 2024
CFP: Due July 14, 2024
Session title options:
From Sensors to Servers: Efficient Edge Computing with Akri and WebAssembly
Event Driven Edge Data Processing with Akri and WebAssembly
Running MQTT Event Driven WebAssembly Applications on the Edge using Akri and SpinKube
Run Lightweight Workloads on Tiny Edge using Akri and SpinKube
Efficient Edge Solutions: Leveraging Akri and SpinKube for IoT and Wasm
Enhancing Edge Applications: Discovering IoT Devices and Deploying Wasm with Akri and SpinKube
From Sensors to Servers: Efficient Edge Computing with Akri and WebAssembly
Connecting Tiny Devices: Edge Data Processing with Akri and SpinKube
Description
The edge is an extremely resource-constrained space as there are hardware limitations with compute and memory. Reducing the footprint of workloads running at the edge is critical for running scalable operations. On top of the resource constraints, at the tiny edge (sensors, cameras, actuators, etc.) there is a heterogeneous ecosystem of IoT leaf devices with different protocols and requirements.
In this talk, we will go over how you can leverage two CNCF projects to optimize your workloads with tiny edge by running serverless WebAssembly applications. Akri dynamically bridges your IoT leaf devices to your edge cluster and SpinKube orchestrates and deploys the Wasm workload on your Kubernetes cluster. We will demonstrate a real-life use case using
Kate's draft
~~At the edge, there are thousands of sensors publishing data that nearby servers process and forward. Workloads processing the data at the edge must have a small footprint as they oftentimes run on resource constrained servers. There are two challenges here at the edge. Firstly, discovering and connecting to the multitude of sensors and, secondly, deploying lightweight applications to the edge that can consume local data. The open source Akri and SpinKube projects come together to provide a faster and smoother experience to connecting to tiny devices from Kubernetes clusters on the edge. Akri discovers IoT devices (sensors, camera, actuators, etc.) and automatically deploys Pods that consume their data. SpinKube enables running serverless WebAssembly (Wasm) applications as Pods on Kubernetes. Since WebAssembly has small binary sizes and sub-millisecond start up times it is the perfect technology for edge applications.
This presentation will conclude with a demo that brings all these concepts and projects together. Akri will discover MQTT based devices and automatically deploy Wasm applications to the cluster. These Wasm applications will be triggered by new MQTT messages, ensuring that data processing applications are only run when needed. By the end of the talk, you will have the tools you need to start running efficient data processing applications at the edge.
Shortened to 1000 words:
At the edge, thousands of sensors publish data. Servers on the edge must discover these devices and then continually process new data from them. Akri and SpinKube are open source projects that address these challenges, providing a seamless experience for connecting IoT devices to Kubernetes clusters at the edge. Akri discovers these devices and automatically deploys Pods to consume their data. SpinKube runs serverless WebAssembly (Wasm) applications as Pods on Kubernetes. Since Wasm has small binary sizes and sub-millisecond start up times it's the ideal application type for resource constrained servers at the edge. This presentation will conclude with a demo of using Akri to discover MQTT-based devices and automatically deploy Wasm applications to the cluster. The deployed applications will be event-driven and triggered by new MQTT messages, ensuring precise resource usage. By the end of the talk, you'll have the tools to run efficient data processing applications at the edge.
Chalkboard/notes
We will show how you can easily install Akri to your edge cluster to start discovering messages from MQTT based devices. A
Wasm = light weight. submillisecond start up times so can execute the processing apps in response to a data publication event. Demo Akri + MQTT + SpinKube – demonstrate how to deploy spin applications as Akri brokers to respond to new messages on MQTT topics.
CNCF Co-Located Event
Kubernetes on Edge Day
Session format
Session Presentation
Level
Intermediate
Benefits to the ecosystem
The audience will learn how to optimize their operations by building event-driven applications for low latency at the edge using open-source and CNCF projects. IoT leaf devices are often too small, old and locked down to run Kubernetes themselves. Akri bridges this gap between your edge cluster and IoT leaf devices by discovering them and creating resources on your cluster to represent them. SpinKube allows you to run light and portable WebAssembly applications on Kubernetes. With this flexibilty, optimizing your edge operations and reducing the footprint has never been easier.
Case study?
No
Presented this talk before?
No
CNCF-hosted software
Akri, containerd, K3s/K8s
Open-source projects
SpinKube, Spin, MQTT, Wasm
Additional resources
Additional notes
WASM I/O CFP
Description
At the edge, thousands of sensors publish data. Servers on the edge must discover these devices and then continually process new data from them. Akri and SpinKube are open source projects that address these challenges, providing a seamless experience for connecting IoT devices to Kubernetes clusters at the edge. Akri discovers these devices and automatically deploys Pods to consume their data. SpinKube runs serverless WebAssembly (Wasm) applications as Pods on Kubernetes. Since Wasm has small binary sizes and sub-millisecond start up times it's the ideal application type for resource constrained servers at the edge. This presentation will conclude with a demo of using Akri to discover a USB-based microphone and automatically deploy Wasm applications to the cluster, which will run an LLM to detect the use of keywords spoken into the microphone such as commands. By the end of the talk, you'll have the tools to run efficient data processing applications at the edge.
