FFmpeg, as a Swiss army knife for open-source audio and video processing, is very popular because of its free openess, powerful functions, convenience and ease of use. The high computational complexity of audio and video processing makes performance acceleration an eternal theme of FFmpeg development. AliCloud video cloud media processing system extensively draws on the experience of open-source FFmpeg in performance acceleration. At the same time, it optimizes the endpoint-cloud integrated media system according to the needs of its own products and architecture, to create a high-performance, high-definition and low-latency endpoint-cloud collaborative real-time media service.

-Rapid growth of domestic audio and video business -Continuous updating and improvement 
-The performance is three times that of similar projects
 -In the background of Internet cloud services 
-Positive cycle: open-source, application, training 
-Accurate positioning of no to do everything

The performance problem in multimedia field is one of the enduring topics. In the cloud, except the well-known distributed processing of task fragmentation, it is also very important for the way of speeding up audio and video processing in more details to customize various workflows of FFmpeg according to business forms. This topic takes the design of cloud multimedia processing architecture as the introduction point, gradually unfolds and shares the performance problems encountered in details, then gives solutions to greatly improve performance, so that listeners can have a more comprehensive understanding of our work in optimizing high-performance FFmpeg applications.

Due to the COVID-19 pandemic, audio and video applications have become indispensable partners in people's work and life. Although many big companies have built video cloud services to cope with the changes in market demand, most companies and developers will choose cloud services to build their own online audio and video applications. This speech wants to give you a clear process of building audio and video capabilities by introducing the architecture evolution of QiNiu Cloud Video Cloud Service. Based on the improvement of user experience, this lecture focuses on the evolution of cloud service process from live broadcast technology to RTC, and how to make audio and video higher (resolution), faster (low latency) and stronger (powerful service capability) step by step. At the same time, the speech will involve the comparison and use experience of some open source technologies in the audio and video field, which will also bring good benefits to companies and developers who need to build their own audio and video services.

1. FreeSWITCH is an open source real-time audio and video communication system, which is not as general and popular as FFmpeg, but still be active like FFmpeg in the communication field. 2. From the end of 2007 when started to contact FreeSWITCH, slowly went from submitting Bug reports and patches to becoming FreeSWITCH core Committer, submitting 140,000 lines of code in 14 years… What kind of experience is it? 3. From FreeSWITCH technology blog to the hand-written “FreeSWITCH Authoritative Guide”, what has this book brought me? 4. What pitfalls have you faced from open source evangelism to getting the first consulting fee, to commercial technical support and products/services? 5. The future can be expected. How to judge the future of open source?

OpenJS Foundation is the global core community for open source JavaScript and other Web technologies, with more than 30 core projects, including Appium, Dojo, Electron, jQuery, Node.js, and webpack. Robin Ginn, Executive Director of OpenJS Foundation, will talk about the latest development of this lifeful foundation, the dynamics of its members, and the planning and prospect of the future open source community in China in this keynote of GOTC.

The scale of front-end code continues to increase, large-scale Web systems, security and compliance all bring increasingly difficult challenges. How to build a standardized, high-quality, efficient, safe and compliant automated front-end engineering system for the delivery of Web systems in large-scale front-end product, to meet higher and higher project requirements, and to guide the team to continuously improve and optimize through data visualization in the R&D process, has become a problem currently encountered by the industry. We will bring DEVUI's enlightenment this time.

At present, cross-end has become an unavoidable topic in the front-end. With the popularity of IoT devices, more and more "ends" emerge in our daily development. Since Hybrid and React Native (Weex), Flutter has become the new favorite of cross-end in the past two years with its simplified rendering pipeline and self-drawing rendering features. However, the development mode of Dart + Widget and the ecosystem separated from the front end have caused the R&D cost to be too high. Based on this background, we explore a Flutter application scheme based on W3C standard, which connects the front-end ecosystem upwards and ensures multi-end consistency through self-drawing downwards.

Before ESM appeared, developers would package to build Web App because browsers lacked JS modularization mechanism. During this period, packaging tools such as WebPack were widely used. However, with the increase of project files, the running time of Dev Server and HMR becomes longer, which greatly damages the development efficiency and experience of developers.
 Vite is a tool created to improve the development experience of developers. It has faster service startup, and a lightweight and faster hot-fix mechanism. This topic will analyze Vite, so that everyone can know more about Vite and be more handy in actual use.

As a basic platform, Fast App platform has been developing rapidly. To support a wider range of scenarios, connect more smart devices, and serve more developers, the Fast App Framework plans to open source for the public, inviting more developers to participate and contribute to the construction of the existing Fast App community. After open sourced, the latest version of the Fast App Framework provides perfect screen size layout support, multi-terminal adaptation, multi-modal interactive integration and other functions, and will continue to bring more and more practical new features in the future.

Cross-end framework liberates front-end productivity, while Serverless liberates back-end operation productivity. And when the cross-end framework meets Serverless, what kind of fantastic reaction will occur?
 This time I will share the cross-cloud and cross-end Serverless solution created by the uni-app team and a series of explorations of uni-app open-source framework in the Vue 3.0 upgrading. The main contents include:
 1. Introduction of DCloud Company and uni-app framework, open-source and commercialization
 2. Cloud collaborative R&D mode
 3. Intelligently generate cloud integrated code
 4. Exploration and practice in the process of Vue 3.0 
upgrading in uni-app 5. Vue 3.0 SSR support based-on Serverless

Type system was a concept with very low presence in the front-end field. But with the rise and popularity of TypeScript, type system gradually entered the field of vision of front-end developers. The improvement in error detection capability, abstraction capability and security brought by it are exactly what modern Web development urgently needs. However, the emerging TypeScript has not solved the type safety problem in all scenarios. In HTTP requests, RPC, cross-end interface calls and other scenarios, there will still be situations where type safety cannot be achieved. Farrow is committed to solve the problem of type safety in server-side development and the problem of type synchronization between client and server, and provides an end-to-end type safety scheme based on TypeScript. This sharing will deeply analyze its implementation principle.

Apache ECharts 5 brings new features such as state management, label upgrade, dynamic narrative, theme visual optimization, universal fillet effect, dashboard upgrade, timeline upgrade, prompt box upgrade, sticker pattern, data conversion, performance improvement, etc. These features focus on visual design, interactive ability, development experience, accessibility and dynamic narration, which ECharts continuously pays attention to. This speech will first show the new features, then share the implementation principles of some features, including state transition design with animation, dirty rectangle rendering principle and broken line graph performance optimization. It is expected that these practices can bring some reference to developers who encounter similar problems.

上海市政府副秘书长 陈鸣波致辞

国家信息安全中心副主任 董大健 致辞

Updating

Jim's career spans three of the biggest technology trends that have emerged over the past decade: mobile computing, cloud computing, and open source software. Today, as executive director of the Linux Foundation, he uses that experience to accelerate technological innovation through the use of open source and Linux. At the Linux Foundation, Jim works with the world's largest technology companies, including Baidu, Huawei, IBM, Intel, Google, Samsung, Qualcomm and others, to help define the future of software technology on servers, clouds and various mobile computing devices. His work at the vendor-neutral Linux Foundation gives him a unique integrated perspective on the global technology industry.

This year marks the 30th anniversary of the birth of Linux. Linux has expanded from an open source operating system project to countless different open source projects today. From supercomputers to cloud computing platforms to smart phones, Linux's influence can be said to be everywhere. Today's enterprises rely more than ever on providing value to customers through open source! In this lecture, Linus Torvalds, the father of Linux, will review the unforgettable experiences on this journey, and share the challenges faced by Linux and the future prospects. At the same time, Linus will also share valuable open source experience and suggestions with Chinese developers and enterprises.

Updating

Updating

Updating

The wind of open source has prevailed. With the rapid upwelling of this hot wind, the open source ecosystem seems to be getting a new life. What is the difference between today's open source and the past? What will happen to open source in the future? How does open source prosper from different perspectives, in different regions and in different fields? How does the open source collaborative model penetrate into other fields today? How does the conflict between open source and cloud providers? How do open source authors survive? How is open source influenced by geopolitics? How does China actively participate in open source? How to commercialize open source? ..... Leaders from several top international open source foundations will bring an in-depth roundtable talk about the new open source ecosystem.

Updating

Updating

Updating

Updating

Updating

Updating

Updating

Updating

Updating

Harbor is an enterprise open source mirror warehouse that can store, sign and scan the content of cloud native products. Harbor project is the first CNCF open source project in China, especially welcomed by Chinese users. According to recent statistics, 47% of users in China use Harbor in production systems. After Harbor graduated from CNCF in June 2020, many remarkable functions have been added, such as mirror agent, online garbage collection, enhancement of robot account, P2P prewarming, performance data collection interface, IPv6 support, etc. Recently, Harbor Operator 1.0 has been released, which greatly improves the usability of Harbor on Kubernetes platform. The main purpose of this speech is to introduce the new functions that Harbor users are most concerned about and the future development direction.

As the only incubated edge computing project under CNCF, KubeEdge has been widely used in Internet, industry, energy, parks, transportation and other industries.
 As the only incubated edge computing project under CNCF, KubeEdge has been widely used in Internet, industry, energy, parks, transportation and other industries.
 This sharing will introduce how KubeEdge can realize cloud-edge collaboration in a low-quality network environment, and realize stable support and management of massive edge nodes and devices.

Alibaba's open source chaos engineering project ChaosBlade has entered CNCF Sandbox, become the first chaos engineering platform project to support the whole lifecycle of cloud native, aiming to help enterprises solve the problem of high availability in the process of cloud native through chaos engineering. This sharing focuses on the positioning of ChaosBlade, introduces the development history and future planning of the project, and shares the commercialization exploration based on this project.

SODA Foundation mission is to foster an ecosystem of open source data management and storage software for data autonomy. In this talk, we will introduce the Open Data Framework project and how it simplifies Kubernetes data and storage management. We will also share how you can contribute to SODA, as well as how your project can join the SODA ecosystem

As we all know, distributed system is very complex, faults are everywhere. So, how to survive in this complex world of distributed systems, especially today, when cloud native architecture can be seen everywhere, has been an actual challenge. And now a good answer is chaos engineering. This speech mainly talks about the technical details of Chaos Mesh, a chaos engineering platform under the cloud native architecture, and various designs and tradeoffs in chaos engineering platform building. It shows the mutual promotion between cloud native architecture and chaos engineering, achieving the effect of 1 + 1 > 2.

Cloud native technology has brought a series of benefits to enterprises, such as cloud infrastructure, Internet core technology, application data and intelligence. With the practice and promotion of cloud native technology, more and more cloud-based enterprises enjoy the dividends brought by cloud native technology. But under the cloud native architecture, Microservices, containers and cloud native components have spawned many new security problems, whole traditional security means cannot be involved. Traditional security protection, endpoints, networks and boundaries are relatively clear at all levels. However, in the cloud environment, these boundaries become very blurred. More importantly, traditional security tools are not cloud native and cannot match the demands of cloud native for flexibility, lightweight, high efficiency, scalability nor cannot be started and stopped at any time. As a result, the efficiency of security protection cannot keep up with the speed of application iteration, and security defense cannot start and stop at any time with the dynamic expansion of services. Therefore, the key requirements of cloud native environment for security construction in management and deployment have changed:
 -Cloud native security needs to cover the depth and breadth of cloud native environment horizontally and vertically from multiple dimensions such as infrastructure layer security, application layer security and data layer security; -Security must be lightweight, continuous and embedded in all aspects of deployment tools to ensure "inspection and control points"; -Security needs to be adapted to the configuration complexity brought by cloud applications, and needs to be automated like applications.

Updating

Cloud native application workloads are well known from Kubernetes native workloads (Deployment, StatefulSet), but on the other hand, we also see that from small and medium-sized start-ups to large Internet companies, the more large-scale application scenarios are, the more these native workloads cannot meet complex business deployment demands. Therefore, many companies have developed their own custom workload suitable for their own scenarios, but among them, only AliCloud’s open-sourced OpenKruise has become a CNCF Sandbox Project, which is truly mature in many aspects such as generalization, comprehensiveness and stability. In this sharing, we will start with Kubernetes native workloads to introduce the responsibilities and implementation basis of cloud native application workload, and then analyze the real demands for application workload in ultra-large-scale business scenarios, how OpenKruise meets these demands, and the development trends under the subsequent open source ecosystem.

Cloud native has become a general trend, but for enterprise customers, for the sake of data sovereignty and security privacy, enterprise customers will consider using multi-cloud hybrid environments to conduct business. However, the differences in infrastructure capabilities and security architectures of different cloud environments will lead to a serious separation of enterprise IT architecture and operation system, increasing the complexity of multi-cloud hybrid implementation and increasing operation costs. This sharing will introduce the development process, typical stages and representative technologies of cloud native multi-cloud & multi-cluster technology, and share how Karmada, the next generation open source multi-cloud container orchestration engine for cloud native, supports seamless migration of applications from single cluster to multi-cluster through native API.

Knative Eventing is a standardized event platform on open source Kubernetes, which provides a set of mechanisms to manage cloud computing event publishing and subscription, and provides event-based service triggers for Serverless platforms. Knative Eventing provides composable primitives to implement binding between event producers and event consumers. This sharing first introduces the overall architecture and underlaying implementation of the Knative Eventing system; Secondly, the event transmission mechanism in Eventing system is introduced. Finally, the application practice of Eventing system in mobile cloud function calculation is introduced.

Service Grid is the hottest technology in the cloud native era after Docker and Kubernetes. It has a great tendency to replace the traditional microservice framework and reconstruct the microservice development mode. Istio is currently the most advanced technology and is also the service grid technology most adopted by Chinese developers. Istio mainly provides rich service governance, security and observability in a non-intrusive way. Service governance and observability may be discussed more in various early technical conferences. This speech mainly focuses on how Istio can help developers build zero-trust network security. The contents mainly cover: automatic encryption of inter-service communication, issuance of certificates, automatic rotation, and authentication related functions.

Volcano is the first container batch computing project based on Kubernetes under CNCF. It provides a set of perfect working mechanisms to solve the scheduling and resource management of high-performance computing services such as AI, Big Data, scientific computing and rendering in cloud native environment. This speech mainly shares Volcano's technological evolution roadmap and typical use cases.

Vipshop’s AI platform supports a complete set of processes and systems of algorithm models from development, training to launch. Its core work includes large-scale distributed model training and model inference optimization. It supports various machine learning and deep learning scenarios such as recommendation, search, advertisement, image and NLP in business to achieve the overall goal of algorithm-driven company performance. This sharing will introduce how Vipshop’s AI platform that relies on Volcano can solve scheduling problems in the training process of TensorFlow model based on K8S: including Gang Scheduling, resource queue management and other practices, and has achieved a good goal of improving resource utilization rate on the whole. In addition, it will also share that Vipshop will use VPA (Vertical Pod Autoscaler) on K8S to dynamically control the use of container resources, that is, to ensure the availability of physical machines and realize the improvement of the overall resource utilization rate under the condition of CPU resources oversold.

GitOps is based on declarative systems (such as Kubernetes) and GitHub (or SCM tools such as GitLab) as a single trusted source, which largely solves the deployment of cloud native applications. However, how to handle the storage of sensitive information on GitHub, how to handle the connection of CI/CD, and how to embed other security contents into GitOps are a series of problems to promote GitOps. This speech will share tekton + kustize + sops + ArgoCD to find a practical path for GitSecOps.

The founder of Docker once said on Twitter that if WebAssembly and WASI were born in 2008, there would be no need to create Docker. WebAssembly is the future of Cloud Computing. It has been three years since this tweet, with the introduction of WASI and the emergence of more and more WASM Runtimes, WebAssembly has made remarkable progress in the development of cloud native field. In this lecture, we will profile several examples of WebAssembly runtime computing at the edge to answer whether WebAssembly can replace Docker in some specific scenarios.

K8S has become a de facto standard and is widely used in production environments. However, due to the problems of underlying technologies and usage methods, enterprises are facing some difficulties in using K8S in the production environment. This speech will base on the experience in the production environment to share you the K8S security dilemma and countermeasures.

eBPF has been regarded as a revolutionary technology of Linux in recent years, which has gradually attracted people's attention and application, and has brought many new possibilities to cloud native scenarios. This sharing will introduce the practice of eBPF in the process of MEGVII PaaS platform construction.