On June 9, 2021, the Asia-Pacific Content Distribution Conference and CDN Summit was held in Beijing. Zhang Yiping, senior technical expert of Ali Cloud Edge Cloud Network, was invited to attend the conference and shared the practice of Ali Cloud in building the accelerated network system of all edge cloud stations, which supported the layered transmission acceleration of various application protocols based on the edge cloud nodes. To provide a wide range of network acceleration infrastructure services for differentiated enterprise applications.
Zhang Yiping believes that speed is a widely concerned business indicator in Internet business, and speed produces value. In the scene of video and interactive interaction, the speed corresponding to the video second opening rate, playback lag rate, video clarity, interaction delay directly affect user experience and retention; Under the transaction scenario of finance and e-commerce, the relationship between request response speed and success rate of each transaction is high value business operation; Under the whole station business docking scenario, one-stop accelerated docking can greatly improve the efficiency of business docking is the key.
Acceleration from content to total site acceleration
Total site acceleration comes from content acceleration. Content acceleration is primarily static content acceleration, such as video on demand, file downloads. CDN improves user request efficiency and bandwidth carrying capacity through a large number of nodes, and accelerates based on cache. In addition to video-on-demand, file download, static web pages and other services that can be cached for a long time, a large number of real-time services appear, which need to rely more on the real network acceleration than the experience acceleration brought by the nearby static cache, such as online games, financial transactions, real-time audio and video communication, etc.
Unified routing scheduling based on edge cloud nodes is the key
The essence of routing scheduling is to achieve global performance/cost optimization in differentiated customer business/scenarios/resources. Resource is node, network, a single node contains CPU computing power, memory capacity, storage IO, network IO, etc. The business is long video, short video, downloads, live streaming, financial trading, edge rendering, etc. Scenario includes cache content (consumption of storage capacity /IO, network IO), dynamic content (consumption of network IO, CPU power), traffic calculation (consumption of CPU power, memory capacity); Routing scheduling can be understood as a process of real-time dynamic solution. On the premise of guaranteeing the quality of service, every type of resource can be uniformly consumed as far as possible, especially the effective utilization of CPU computing power resources of edge cloud nodes.
Intelligent protocol stack is the basis of improving service quality
Combined with the difference of service, scene and environment, the more suitable transport protocol stack can reduce the service difference of different resources. In the scenario of protocol stack optimization, it is divided into terminal link, intermediate link and back link.
1) Terminal link
Terminal oriented, make full use of the data difference of terminal environment, drive and match the most suitable transmission algorithm. The difference of network environment, service type, time and other aspects during terminal access reflects different service quality. Transmission protocol matching is the most effective means to optimize the quality and also optimize the network cost.
2) Intermediate link and back link
Large-scale link convergence, the pursuit of optimal transmission quality and stability, allowing the sacrifice of part of the cost, against the public network transmission fluctuations and interference, to achieve disaster tolerance link.
Full link diagnosis is the cornerstone of fast fault location and recovery
!
From the perspective of business, it can quickly detect and locate abnormal points, improve the recovery speed of abnormal points and improve the efficiency of operation and maintenance. It can be divided into two types of detection:
1) Passive detection
Including TCP marking data statistics, starting from the seventh layer of business, marking tracking based on the Request/Response dimension of the business layer, packet dyeing according to the seventh layer of business marking when transferring from the seventh layer to the fourth layer, statistical reporting of dyeing packets of various gateways of the intermediate link, and finally diagnosis of abnormal points based on the statistical data.
2) Active detection
With the help of a large number of terminals to feedback the service quality from the end user’s perspective, through its own thousands of nodes of fine mutual detection, linkage abnormal automatic disposal, in addition to the third party broadcast platform can be used to obtain the third party perspective of service quality evaluation.
Finally, the original three/four/seven layers of independent quality detection is upgraded to three/four/seven layers of penetration correlation, forming a global perspective of accelerating the quality of the whole link at all stations.
Build a systematic security protection capability
When talking about network security, the first thing that comes to mind is traffic attacks such as DDoS. As a basic ability, this must be required. It is a resource-based ability. Ali cloud has enough resource pool to prevent traffic attack.
Zhang Yiping pointed out that in addition to resource side protection, it is also necessary to build a security system, especially for dynamic content. Dynamic accelerated content has the characteristics of small request bandwidth and large business value. Compared with the most basic violent traffic attack, dynamic content is more about crawler management, business brush prevention and other functions. For example, all kinds of prices and orders on e-commerce websites are a lot of dynamic content data, which will face more crawlers and order brushing, etc., and the protection that needs to be done is also more for these. Through establishing a complete set of security system of data analysis, do in-depth cooperation in business level, with the customer together to define what kind of dynamic request is normal, what is abnormal, and the abnormal request of different business what are the characteristics, how to respond to as far as possible to improve security ability and decrease of hitting probability;
The establishment of a security system depends on the accumulation of offensive and defensive techniques and strategies, and this is the most valuable.
The core capability of network acceleration is also the foundation for building the edge cloud system. In each edge application scenario, the acceleration of the underlying network and the interconnection between multi-clouds can help build the edge cloud ecology, interface standardization, cross-network collaboration and cross-cloud business deployment, and provide broader and universal infrastructure capabilities for the upper layer of business.
At the end of the speech, Zhang Yiping expressed her future-oriented views: in the development of edge cloud ecology, scattered delivery of basic resources will become more and more difficult to implement, scene application solution delivery will be the mainstream direction, and open cooperation and win-win will be the main theme. She is looking forward to strengthening industrial cooperation and creating edge cloud ecology.