Open Source Summit Japan & Automotive Linux Summit 2021

This abstract is about Centaurus an open-source project that brings a next-generation cloud for Edge and 5G. We are driving innovation across multiple industries to offers high scalability for large cloud infrastructure. 5G and edge computing are two inextricably linked technologies: they are both poised to significantly improve the performance of applications and enable huge amounts of data to be processed in real-time. Under Linux foundation, Centaurus is an open-source project that is driving the top industry requirements for Edge and 5G. We would like to introduce Centaurus an open-source solution for large-scale new cloud infrastructure. It helps to seamlessly extend from cloud to edge, and natively manage edge workloads from the cloud. Edge computing matters because it increases the speed and effectiveness at which businesses can respond to information. • Edge computing and 5G are increasingly being deployed. Centaurus platform intelligently coordinates resource allocation between cloud data centers and edge sites. • Our design aims to support 300,000 hosts per region and 100,000 hosts per cluster. • Application orchestration is the norm in the container world — such as replica set, etc. Centaurus can orchestration paradigms also be applied to VM applications.

Have you ever mistakenly deleted Kubernetes resources? Did it happen in a production environment? Could you successfully recover from it? If you have such an experience or if you think that it could be your real-world problem in the future, this session is for you. This session will show you how the Kubernetes community has worked on to avoid such issues for PersistentVolume and PersistentVolumeClaim, and how the community is trying to solve such issues in general, along with protecting Secret particularly. The use cases of the protection in this session cover both deletion orders caused by resource dependencies and protecting resources that users marked as important. Some of the features explained in this session are still under development. However, both how you can configure with the current features and how you will be able to configure with the new features will be explained with Pros and Cons. So, you will be able to start using some of what you learned, right after this session!

Most organizations have some kind of legacy systems. Even after they introduced Kubernetes, they usually don't migrate everything to Kubernetes. As a result, connectivity between legacy systems and Kubernetes sometimes makes problems. One of such problems is how to connect legacy systems that identify senders by using source IPs and Kubernetes that doesn't provide stable source IPs. This session focuses on this problem and explains (1) what kind of legacy systems requires identifying senders by using source IPs, (2) why the Kubernetes network model and Kubernetes features don't provide stable source IPs, (3) How submariner, which is a tool to provide multi-cluster connectivities, helps solving this problem, and (4) Current status of the solution in the community.

Traditional CI/CD systems haven't been designed for cloud native environments and need to evolve. Tekton is a flexible, Kubernetes native open source CI/CD framework that enables automating deployments across multiple platforms—including Kubernetes, serverless, and VMs—by abstracting away the underlying details. It provides a set of shared, open-source components for building CI/CD systems and is one of the founding projects of the CD Foundation (CDF). Within the CDF’s open, vendor-neutral structure, Tekton’s goal is to provide industry specifications for CI/CD pipelines, workflows and other building blocks. This talk explains the CI/CD challenges in a cloud native landscape shows how Tekton project addresses these challenges. It goes into details on what Tekton resources are, how to install the pipelines and how to run them for building, testing and deploying containerized applications on their clusters.

In this presentation, Shukun will introduce a new way for bare metal management which is automatically manage bare metal hosts and related network devices via a Kubernetes (k8s) style API, so that in hybrid cloud environment, on-premises infrastructure can be managed like cloud. In hybrid cloud environment, create and manage the on-premises side infrastructure is very tedious as the operator may need to not only deploy the whole environment, but also configure network for isolation, configure RAID and BIOS for reliability and optimization and so on. It is necessary to automate such operations so that on-premises can be treated just like cloud. K8s is a great tool for managing compute resources and there are many related projects which can extend the scope of k8s, such as: Cluster-API and Metal3.io. These projects can be used to achieve the above goal, automate on-premises management operations. In this presentation, Shukun will explain these projects, demonstrate how to use them and introduce the latest development happened in the community. Audience could understand how to automatically manage bare metal hosts and the related network devices through a k8s style API, and this could help to extend the use of k8s and finally lead to further prosperity of the community.

The traditional computer architectures are no longer able to keep pace with ever-growing demand. The future computing infrastructure needs to be dynamic, accelerated and secure in nature, just like the containerized software today is composable and on-demand. Cloud platform is expanding to the edge of the network while supporting data and AI workloads requires new approaches to hardware. Network workloads such as Open vRAN also require new approaches to hardware in order to replace traditional vendor specific hardware. System architectures are transitioning away from the general purpose computing model. Adding more CPU is not sufficient and offloading with traditional hardware acceleration approaches cannot meet the demand. Some communities such as IOWN Global Forum already started to address disaggregated and heterogeneous computing with common hardware chassis. This session elaborates system architecture innovation in open source and how kubernetes and Linux with heterogeneous composable computing are enabling these exciting new capabilities in disaggregated infrastructure. We will discuss vision, use cases, architectural approach, and future with valid examples.