 Hello everyone. I'm glad to share with you at Kubicon. I'm Huang Jian, the Chief Architect of Huawei Cloud Container Service Team. I've been a member of the team of developing cloud-native services since 2013. I've experienced the ups and downs of the cloud-native transformation. Today I want to share how to build an open-source distributed cloud-native platform. We have seen the fast development of cloud-native technologies such as containers and service mesh. Starting from the internet sector, customers are coming from more and more industries such as finance, transportation, manufacturing and energy. They embrace cloud-native and seek for digital transformation. According to IDC, the proportion of cloud-native applications will reach 80% by 2023. As of now, we can find that apart from the upper-layer apps, more and more middleware has been containerized and runs cloud-native. Service software, big data, AI, caches and functions can run in a unified cloud-native resource pool. Cloud-native is there for extending from a single central cloud to edges and devices such as the use case of highway tolling. It's even safe to say that the full stack of what apps need now is to be cloud-native. This year, Huawei Cloud proposes the ubiquitous cloud-native concept and promotes a new model of distributed cloud-native cloud infrastructure. So, what is distributed cloud-native and what are the features? Distributed cloud-native provides unified full life cycle management for cloud, edges and devices. First, unified application management. This involves consistent multi-cloud management, cloud edge device collaboration, unified resource scheduling and flexible management across distributed clouds. Second, unified traffic management. You can centrally manage service meshes running on the cross-domain, cloud edge and fully interconnected network infrastructure. Third, unified data management, including cloud edge storage acceleration, data acceleration and data migration. Data goes wherever your apps run. Finally, unified ONM management, including unified monitoring log view, mesh tracing and outlier detection. All of these will help you better manage your applications. In recent years, Huawei has been polishing the distributed cloud-native platform and contributing its core software to the CNCF community. Huawei has open sourced several projects. In 2018, Huawei launched KubaEdge, which became the first intelligent edge computing project of CNCF. KubaEdge brought cloud-native to the edge. In 2019, we launched Volcano, the first batch computing project of CNCF. Volcano allows big data and AI jobs to smoothly run on cloud-native platforms at scale. In the first half of 2021, Huawei cloud launched Kamada, a distributed cloud container orchestration project. We have been working together with so many contributors to build this software. By now, it's running in many productions. We donated the project to CNCF and it has become an official CNCF project. So how to use these open-source software to build a distributed cloud-native platform? I'd like to share some light on this. First, you can use Kamada to orchestrate multi-cloud containers. Kamada serves as a unified entry and manages multiple Kubernetes clusters across clouds and regions. Kamada is compatible with native Kubernetes APIs and extends the scheduling policies for cross-cloud applications. We keep testing compatibility for mainstream cloud-native software, including ArgoCD and Helm on Kamada. We also want to build a fully interconnected container network. In this regard, we use Submariner to test Kubernetes cross-cloud and cross-cluster networks. Second, you can use KubaEdge to extend your cloud-native platform to the edge to manage your edge software. Widely distributed edge nodes are no longer a headache to manage and you can enjoy a lightweight running environment and offline autonomy. KubaEdge has become ready to use for mobile edge devices. Community members have used it in cars and even satellites. KubaEdge has many updates this year. For example, EdgeMesh realizes cross-edge service-oriented mutual access. The Sedna Edge Cloud AI collaboration suite brings in federated inference and enhanced training of the edge and cloud. The Mapper Edge device suite has also been refactored. As more access protocols are supported, you can now manage more edge devices. Third, you can use Volcano to expand your cluster. It provides diverse scheduling policies to help you better run batch services. Now, Volcano is upgraded to support hybrid service development. Online and offline services can smoothly run in the same cluster. We will soon release Volcano Global in the community. Volcano Global will integrate Commodore and provide intelligent hybrid scheduling of applications across clusters. When applications become more distributed, traffic management becomes more challenging. In this case, you can integrate Istio to improve the traffic governance for multi-cloud and multi-cluster applications. Istio can well support multi-cluster and distributed cloud deployments. You can use a unified Istio control plane to manage your service mesh. The data plane can be as lightweight as Envoy Sidecar or GRPC Sidecar to access the service mesh. We also advise you to use Istio Locality to better manage the traffic. Finally, don't forget the observability of cloud native applications. In a cluster, you can use Prometheus to collect cluster monitoring data. On the edge side, you can use the lightweight agent of OpenTelemetry to collect monitoring data of edge nodes. In the central cloud, you can use open-source software such as Cortex to collect monitoring data and Grafana to check the monitoring data. Besides, you can also use OpenTelemetry to collect mesh traces. Thanks to the multi-cluster feature of Kiali, you can monitor multi-cluster meshes in a unified manner to have a clear overview on your distributed applications. So far, you have built a base for your own open-source distributed cloud native services. Integrating all these open-source capabilities, Huawei launched UCS to inject cloud native into your applications running everywhere. UCS provides intelligent application scheduling and scaling, find grained traffic governance and collaborative management of applications and data so that applications will not be throttled by cross-cloud, cross-region and traffic management. Huawei Cloud will stay open and keep contributing to open-source communities to build a distributed cloud native world. We welcome you all to join us. Thank you.