On June 9, 2021, the Asia-pacific Content Distribution Conference and CDN Summit was held in Beijing. Li Songlin, intelligent edge cloud technology expert of Ali Cloud, was invited to participate in the interactive live forum to share the design ideas, technical principles, characteristics and application practice of global real-time transmission network GRTN based on Ali Cloud edge cloud node. And provide stable and reliable business experience for customers of live streaming applications.
The picture
At present, there are two main application architectures of live streaming technology: direct push and back-source pull. The reasons for these two architectures are relatively simple: First, business scenarios need to connect to the MAC, which requires low-delay cloud convergence; The second is based on UDP private protocol push stream. In order to ensure that anchors can have good streaming effect under the condition of weak network, the most important reason is that mainstream cloud vendors do not have universal mature services for large-scale application of low-delay interactive scenes, so they need to establish their own source sites. Use only the distribution capabilities common to cloud vendors. Supports large-scale distribution of HTTPFLV, RTMP, and HLS.
Pictures of the challenges faced by the self-built live video service
As the scenes and contents of live broadcast become more and more rich and professional, there are more and more demands for interaction, and the delay requirement for interaction is also higher and higher. The original architecture is difficult to meet the demand. Unable to meet their own business needs, many enterprise customers have begun to try to establish their own source stations and optimize protocols. However, due to the relatively high threshold of live broadcast technology, it requires not only investment of resources, but also professional research and development capabilities, as well as long-term continuous operation and management.
Location picture of Ali Cloud GRTN
In order to reduce the end-to-end delay of live broadcasting, Ali Cloud builds a Global real-time transmission Network (GRTN) from application scenarios such as live broadcasting, short delay live broadcasting and RTC. Li Songlin introduced that Ali Cloud GRTN positioning is based on the public cloud center Region and edge cloud nodes, to build ultra-low latency, fully distributed sinking communication level streaming media transmission network. GRTN currently integrates the transmission and exchange of audio and video streams in various business scenarios such as Internet live broadcasting and RTC. RTS with short delay based on GRTN can support the standard H5 WebRTC push and play, and the delay can be controlled within 1s in the case of tens of millions of concurrent levels. The RTC end-to-end delay can be controlled at about 250ms. GRTN can provide three atomic capabilities: publishing, subscribing and switching of streams. Users can build call scenes, live broadcast scenes, link scenes and so on based on these capabilities.
Ali Cloud GRTN architecture picture
The overall architecture of GRTN of Ali Cloud is evolved from the original live broadcasting system. The architecture has the technical characteristics of control and data separation, hybrid networking, multipath transmission, self-learning Qos and so on, and can support multiple access protocols (RTC/RTMP/HLS/HTTPFLV/SRT/QUIC). The core values of GRTN are as follows: Cost reduction. GRTN is a multi-service integrated network that supports multiple scenarios such as live broadcasting, RTC, and video cloud, with high service reuse rate. In addition, GRTN has shorter internal links and lower costs within nodes.
To improve quality, GRTN internal network supports the network structure constructed by dynamic route selection. The internal link delay can reach about 20ms, and the internal link adopts private protocol for efficient transmission. In addition, the push stream and distribution of the client are built based on WebRTC. QoS congestion control is designed specifically for streaming media characteristics, and it is continuously iterated and polished based on online data construction.
Easy to expand, GRTN supports WebRTC protocol, and can carry out full duplex communication on a single connection channel, so that it can publish and subscribe media streams freely, bringing more imagination space in service scalability.
GRTN Key Technology – Distributed heterogeneous deployment image
When talking about the deployment of GRTN, Li Songlin pointed out that the data surface of GRTN can be deployed on different resources to achieve one code and multiple resource deployment, meeting the needs of low delay global coverage. In addition to the original CDN node coverage resources, CDN also supports resources such as central Region and MEC, which improves service experience.
GRTN key technology – peer-to-peer networking and dynamic path planning pictures
Efficient use of abundant resources is the core. Traditional path planning mainly focuses on the quality and takes little account of node attributes and water levels. At the same time, due to the reusability of streaming media, when a stream has already appeared at a node, the whole path choice will face new adjustment. The current strategy is to find a good set of nodes and paths by probing path selection. Create a node path status table. The different dimensions of each path are quantitatively scored. A new optimal solution is obtained by combining weights and strategies. GRTN adopts the hybrid networking mode, that is, the combination of hierarchical structure and peer-to-peer graphics. The uplink selection center periodically collects the results of internal link detection. To coordinate with dynamic networking, the streaming media brain module needs to manage the flow information and support path switching, capacity planning, and comprehensive scheduling between cost and quality.
The key technology of GRTN – two-way real-time message network picture
With the control plane path planning and policy control, how to quickly and accurately deliver data to every node, every machine is also a challenge. In the RTC scenario, a commonly used function is Mobility of the client network. For example, when users go home or leave home during a meeting, the mobile network needs to switch between 4G and wifi. In addition, when the CDN node connected to the client is abnormal, In both cases, the client will switch access nodes in the communication process with GRTN. The two-way real-time message network constructed by GRTN can achieve the millisecond transmission of network cutting messages. When a media stream of a distributor switches over the network, the subscribing client is completely unaware of the switching behavior in GRTN.
GRTN’s key technology – streaming media twin images
According to Li Songlin, GRTN has designed a Streamimg Media Digital Twin system, which is used for capacity evaluation, algorithm training, event replay and analog pressure measurement. By dividing the system into real and virtual environments. In the real environment, real scenarios and data are collected. In the virtual environment, capacity assessment and algorithm training are performed. When the new policy is modified, the new algorithm can be applied to the virtual environment through real historical data input. Through data processing, data reports are generated to compare the previous data in the real environment. This can guide the new algorithm tuning and can also evaluate whether the new algorithm is effective.
GRTN key technology – programmable pictures
The upper business scenarios of media technology are very rich, such as e-commerce live broadcasting, video conferencing, online education, enterprise live broadcasting, new retail, etc., so there are many requirements for customized development. Programmable transformation is an attempt of GRTN to improve system stability. At present, GRTN’s central streaming media brain, node side business module, media data transmission module, media signaling processing module and other modules have been programmable transformation, and binary publishing can be avoided in most cases.
The key technology of GRTN – whole link visual monitoring image
Li songlin believes that observability is the basis for evaluating whether a system can serve the outside world. When online problems occur, we can locate and deal with them in a timely manner to avoid further impact. At the same time, data can also be collected through the observation system to continuously optimize the system.
Create ultra-low delay live RTS pictures based on GRTN
In order to facilitate customers and industries to embrace GRTN, Ali Cloud created an ultra-low delay live broadcast service RTS based on GRTN, which has four technical features: Second delay and excellent anti-weak network capability, the delay can be reduced by 80% at the same lag rate. Compared with the 5-10s delay of traditional RTMP and FLV, the delay of RTS can reach less than 1s, and it is also based on online big data, in self-learning and continuous iteration.
Mature and stable, RTS after more than 2 years of research and development, and experienced taobao live 618 promote online test, has been in Taobao live online.
Open standard, in order to facilitate the development of player customers to use our RTS service, Ali Cloud WebRTC access signaling protocol completely open, transparent.
With wide coverage and high concurrency, THE RTS service is built on Aliyun 2800+ edge nodes, which can support tens of millions of concurrent playback.
Li Songlin also shared specific cases: Taobao Live made the first large-scale use of GRTN technology in the global real-time transmission network on November 11, 2020, and the interactive experience was greatly improved, the conversion rate of transactions was improved, and GMV of live broadcast goods was increased by 5%. Facing the future, more and more live streaming will be applied to people’s lives. Aliyun will continue to increase technological innovation in the field of live streaming to get through the last kilometer of live streaming. Relying on 2800+ edge cloud nodes around the world, Aliyun will provide stable, reliable and safe live streaming services and create live streaming experience closer to real scenes for users.