One cloud, one network, integration-GRTN creates the best streaming media scene practice

Introduction to Cloud's streaming media cloud native platform

“

Alibaba GRTN is designed for streaming media cloud native design, which is convenient for customers to build their own streaming media cloud native applications and make streaming media services ubiquitous.

”

At the recent distributed cloud theme report meeting, Alibaba Cloud senior technical expert Lu Ri delivered a wonderful speech entitled "GRTN to create the best practice of Alibaba Cloud edge cloud streaming media scene".

Lu Ri shared his insights on distributed clouds. He said that Gartner has listed distributed clouds as the top 10 technology trends in the world for two consecutive years, and predicts that by 2025, more than 50% of enterprises will use distributed clouds. Distributed cloud represents the future of cloud computing.

Distributed cloud can be divided into local public cloud, metro area cloud, 5G mobile edge cloud, IoT edge cloud, network edge cloud, etc., serving different user scenarios. Distributed cloud is currently in an innovation-driven period, and the future is full of challenges, opportunities and imagination.

Cloud proposes a cloud-side-device collaboration system. Cloud + device forms a cloud, which coincides with the concept of distributed cloud.

More and more ecological applications rely on cloud-edge-end collaboration, and the edge cloud plays a role as a link between the system: on the one hand, it integrates various edge nodes to build ubiquitous edge coverage; on the other hand, , Based on self-developed distributed operating system, realize unified management and unified scheduling.

Alibaba's next-generation streaming media network GRTN is a typical distributed streaming media practice built under a cloud-side-to-device collaboration system. Lu Ri summarized it in nine words: a cloud, a network, and integration.

A cloud means that Alibaba Cloud GRTN is built on the central cloud native and edge cloud native infrastructure, and organically integrates the technology, draws on the design concept of SDN, separates the CD, puts control in the center, and distributes the data plane. Sink on the more than 2,800 nodes of Alibaba Cloud Edge Cloud.

A network with scene-based QOS capabilities, real-time communication capabilities within 400 milliseconds, and ultra-low latency capabilities. GRTN also has full-link RTC and dynamic networking capabilities. These capabilities are the next-generation streaming media transmission network. Typical characteristics.

Integration, GRTN provides an integrated solution that not only supports the streaming media features of video on the cloud and video distribution, but also has distributed computing and distributed storage processing capabilities. GRTN can provide enterprises with easier-to-use and more professional services for building streaming media applications. Alibaba Cloud has turned GRTN into a cloud native aPaaS service for streaming media, with more agile computing, more real-time connections, more diverse resources, and more refined Control, more scientific evaluation and other features.

More agile computing

A typical streaming media full link is divided into seven links: acquisition, encoding, sending, distributing, receiving, decoding and rendering.

In the traditional architecture, in order to build enterprise streaming media cloud applications, developers usually deploy self-developed streaming media processing programs and software or systems to virtual machines or containers in cloud centers and edge computing. This is to some extent This is not only a waste of resources, but also additional research and development costs to deploy and schedule virtual machine resources.

In fact, what developers really need is a landing environment for media processing software, rather than general facilities that are bundled with operating systems, CPUs, disks, and network cards.

Therefore, Alibaba Cloud advocates that the streaming media cloud native platform provides serverless services for developers for heavy computing. Developers can use self-developed media processing programs. Lu Ri calls them Serverless streaming media operators. The streaming media cloud native platforms can be used on demand Deploy and schedule operators. Operators can be divided into pre-processing operators and post-processing operators, which perform pre-processing and post-processing for the media stream.

More real-time connection

Among the seven links, the middle three links are related to transmission and networking. The core indicators of this sub-field mainly include scale, such as whether to support tens of millions of concurrency and some core QoE indicators, such as first screen, delay, card Wait now.

Lu Ri said that in the future, low latency will inevitably become a standard capability in all walks of life. Therefore, higher requirements will be put forward for the streaming media network networking method.

On the one hand, more node resources need to be expanded in construction; on the other hand, the system needs to strengthen the ability to adapt to various heterogeneous resources. While providing high-quality link resources, it takes into account the stability and easy maintenance of the system.

Alibaba Cloud advocates the use of Hybrid hybrid networking for the streaming media cloud native platform, which can achieve full-scene coverage in the future, and terminal-oriented protocol adaptation for various scenarios.

Therefore, streaming media cloud native platforms have higher requirements for the rapid development and launch of streaming media protocols. Media streaming is the basic unit of streaming media cloud native platform management. Alibaba Cloud provides multiple networking methods, compatible with rich streaming media protocols, While pursuing low latency, pay attention to other indicators of the system. It is necessary to intelligently adapt to these different scenarios and business requirements to build a multi-dimensional streaming media orchestration system.

More diverse resources

In the field of cloud computing, the multi-cloud model has been generally accepted by many companies. Multi-cloud is a major trend, and multi-cloud has produced great value for customers.

When a supplier fails, you can quickly switch to another supplier to minimize the unavailability of the business. Improving business stability is a typical multi-cloud value. Therefore, we can learn from the multi-cloud idea to improve the stability of the cloud service itself. Alibaba Cloud's streaming media cloud native platform can have more diversified resource hosting and scheduling capabilities.

As the technology of containerization matures, the use of containerization technology to layer businesses and isolate tenants can minimize the impact. Although new challenges are raised for software management and operation and maintenance deployment, the improvement of business stability is obvious.

Business innovation and product iteration efficiency can be flexibly controlled depending on the customer's situation, and at the same time, the flexibility of the system can be improved, which has very good business value.

Finer control

Traditional streaming media services actually have many limitations in terms of openness, autonomy, and convenience. As a result, streaming media cloud vendors are channelized and have low innovation capabilities; long links and high costs.

The streaming media cloud native platform can provide more fine-grained control over the platform, including programmable, observable, extensible, and low-code features.

needs to build a variety of controllers, which can be divided into two categories, one is the heavy-computing controller, and the other is the light-computing controller . A typical light computing controller allows third-party customers to extend their own QOS algorithms in a programmable way; the distributed serverless video operator distribution platform can be classified as a heavy-computing controller, and third parties can also implant other operators.

In addition to business isolation, it is also necessary to further adopt secure container technology to further reduce the security risks caused by the opening of the controller. An open and independent strategy can stimulate the innovation of streaming media. In the future, an operator mall can be built to disseminate valuable operators and algorithms in the industry.

More scientific evaluation

Quality optimization is one of the important tasks of streaming media providers. There are some common pain points in the quality optimization process, such as excessive reliance on the experience of tuning engineers. This experience is difficult to precipitate and the cost of trial and error is high.

Alibaba Cloud's streaming media cloud native platform can build a more scientific and closed-loop test system with the help of the concept of A/B testing to achieve the continuous optimization and iteration of QOE. Programmable design can effectively reduce costs and be driven by data. And evaluate the tuning effect, quickly approach the ceiling. In addition, the data platform also needs sand table simulation capabilities, which can make a priori predictions for tuning and reduce online regressions.

GRTN short-latency live broadcast RTS typical case

GRTN has undergone collaborative research and development by multiple teams of Alibaba and polished in a variety of business scenarios. At present, many apps and websites of Alibaba Group have launched low-latency live broadcast capabilities, including Taobao, Dingding, Youku, Lazada, and so on.

It is worth mentioning that compared to the traditional live broadcast user experience, Taobao e-commerce live broadcast delay on Double Eleven 2020 has been reduced by more than 80%, and fluency has increased by more than 30%. This year’s Double Eleven will further compress the delay and improve and clarity 1615bff013562f.

In terms of external commercialization, GRTN provides thousands of domain names to help customers provide live audiovisual experience. For example, after the GRTN RTS live broadcast was launched on a short video platform, the live broadcast latency was reduced by more than 70%, and the video rendering lag was reduced by more than 40%.

Finally, Lu Ri said that he hopes to use the good situation of distributed cloud computing to promote more inclusive streaming media cloud services. While continuing to serve the Internet industry, it will better empower traditional industries and find a balance between generality and convenience.

Alibaba GRTN is not only based on cloud native, but also 1615bff013568f, which is designed for streaming media cloud native design. committed to providing streaming media cloud native services, so that our customers can build their own streaming media cloud native applications and stream media services. everywhere. Alibaba Cloud will work with all colleagues to create an inclusive streaming media cloud native ecosystem to provide users with a more real-time, smoother, and higher-definition audiovisual experience.

---

"Video Cloud Technology" Your most noteworthy audio and video technology public account, pushes practical technical articles from the front line of Alibaba Cloud every week, and exchanges and exchanges with first-class engineers in the audio and video field. The official account backstage reply [Technology] You can join the Alibaba Cloud Video Cloud Product Technology Exchange Group, discuss audio and video technologies with industry leaders, and get more industry latest information.

Copyright Notice: content of this article is contributed spontaneously by Alibaba Cloud real-name registered users. The copyright belongs to the original author. The Alibaba Cloud Developer Community does not own its copyright and does not assume corresponding legal responsibilities. For specific rules, please refer to the "Alibaba Cloud Developer Community User Service Agreement" and the "Alibaba Cloud Developer Community Intellectual Property Protection Guidelines". If you find suspected plagiarism in this community, fill in the infringement complaint form to report it. Once verified, the community will immediately delete the suspected infringing content.