Open Source Summit + AI_dev Japan 2024

This talk explores groundbreaking methods for enhancing cross-platform AI workloads within container ecosystems. The focus will be on integrating the WebGPU standard into containers, enabling them to utilize host GPUs and other AI accelerators through a flexible API. This approach eliminates the need for creating container images specific to individual GPU vendors and their proprietary drivers. Explore 1. How the WasmEdge project leverages the WebGPU standard to develop portable LLM inference applications in Rust. 2. How containers can efficiently manage and orchestrate these applications, simplifying AI deployment across diverse environments.

Security is paramount in today's online world. To enhance security, most small routers have a read only operating system and read write storage for configurations and general storage. Adding functionality usually requires uploading new firmware to the device which can be costly in both data charges and time. Wouldn't it be great to be able to maintain the security of a read only filesystem, but still be able to install new features? Containers are isolated from the host system, maintaining security, but are able to be installed in the read write portion of storage, or even run completely in RAM. The problem with containers is the resource requirements just to run said container. In this talk, we will demonstrate how a minimal, but fully functional container subsystem can be run on a small, resource constrained embedded device. We will show how containers enable edge computing where the user can run their own software on the device without compromising security, or needing to integrate their application with the host operating system. Containers can turn even the meekest of devices into an edge computing powerhouse!

As AI and ML become popular in container (K8s) environment, enormous computational resources are required more and more. On the other hand, efforts toward energy efficiency are also required for the realization of a sustainable society. It is expected to achieve the conflicting requirements that providing higher performance and reducing power consumption simultaneously. Recently, a new server architecture called Composable Disaggregated Infrastructure (CDI) is emerged. CDI can provide custom made servers by composing devices such as compute, memory, storage and GPU connected to PCIe or CXL switch fabric as a resource pool on demand. CDI can provide appropriate type and number of devices on demand depending on container workloads, therefore it would be a solution for this requirement. However, there is currently no standard way to control CDI from K8s. In this talk, we show how K8s and CDI could be a solution. We provide CDI operator and its Custom Resource Definition. Also, we introduce advanced vertical and horizontal cluster auto scaler for CDI. We demonstrate how K8s dynamically attach or detach devices to nodes combined with Dynamic Resource Allocation (DRA) function.

Imagine a world where the hassles of configuring dependencies disappear, replaced by a streamlined and automated process. Buildpacks enable developers to focus on writing code instead of dealing with complex configurations. This session will unveil the secrets of Buildpacks, highlighting their capacity to seamlessly manage various languages, frameworks, and runtime environments. Discover how Buildpacks revolutionize your development pipeline, fostering consistency and efficiency. A sneak peek to our session: *Core principles & benefits of Buildpacks *Application packaging/deployment *Using Buildpacks to enhance collab across development teams Whether you're a seasoned CI/CD practitioner or a DevOps enthusiast, this session equips you with the knowledge to transform software development with Buildpacks. Join us to explore the exciting future and the impact of Buildpacks on the CI/CD landscape.

Container images are ubiquitous nowadays, you've probably used them in one way or another or might have even built one of them. But how are they constructed? What's inside of one? This talk goes through the process of manually creating a Container image without using tools such as docker. Along the way, we'll cover essential components such as layers, metadata, and configuration options that make up an image, and use them to assemble our final Container image. We'll also load it into Docker and run it! By guiding you through the step-by-step assembly of a Docker image, this talk aims to demystify Docker images and strengthen your understanding of their inner workings.

Today, many companies are using Kubernetes in production environments, but there are various challenges in commercial use. One of the challenges is managing the Kubernetes cluster. Many companies are facing challenges related to Kubernetes version upgrades. According to a 2022 survey, most host systems run Kubernetes versions that are 18 months or more out of support. In this situation, the Kubernetes community launched the WG-LTS (Long Term Support Working Group) to promote Kubernetes LTS (Long Term Support). In this session, I will introduce the background, purpose, and activities of the WG-LTS. I will also explain why as a System Integrator, NEC needs Kubernetes LTS. On the other hand, (This is not just in case of Kubernetes, but) there is gap between community developers and users in open source community. I will discuss such gap also. I will also discuss the barriers(human resources, cost, etc…) to advancing LTS in Kubernetes community.