Kingsoft Cloud Zong Jie: Face CDN pipeline, build CDN based on edge computing

The "2019 Asia-Pacific Content Distribution Conference" and the "2019 Asia-Pacific CDN Summit" were officially held in Beijing on April 10. Zong Jie, general manager of Kingsoft Cloud CDN and Video Cloud Product Center, attended the conference and gave live speeches on hot topics such as CDN, edge computing and AI to explore technology trends and prospects.

With the deployment of 5G, AI and IoT work together as double wings. In the information age, everything connects to the Internet. The Internet has entered a new phase in which the CDN architecture has been unable to meet the application needs of the 5G era. Therefore, CDN will usher in the third stage of development characterized by edge computing plus AI. Faced with complex application-side and low-latency requirements, CDN needs to respond quickly and achieve greater transparency in service capabilities, status and quality. What changes will happen to the future CDN model? What kind of sparks will there be in the combination of edge computing and CDN?

With these questions, the reporter invited Zong Jie at the meeting to exchange views on the industry changes brought about by the current "edge computing" and the future development of CDN.

Deep cultivation, making traditional CDN develop into refined management

The first unavoidable question is whether the CDN business is currently making money. Judging from this problem alone, Zong said that this is a very interesting phenomenon. A large number of players flocked in a short period of time, causing excessive competition, resulting in a rapid decline in gross profit. On the one hand, the gross profit margin is constantly being compressed, while on the other hand, new players are constantly pouring in, and a large number of companies have to seize market share even if they lose money. After years of evolution, CDN has formed an industry and become an indispensable basic service on the Internet. It is an indisputable fact. It is precisely because of the future development direction of CDN that many large enterprises, such as Kingsoft Cloud, are deeply thinking about how to develop more business to obtain profit on the basis of CDN.

How to improve profitability? General Manager Zong Jie believes that the extensive operation mode should be gradually transformed into refined operation, which includes two aspects. The first is to enhance market competitiveness and expand market share under the conditions of continuous improvement of service quality and service stability. Second, fine management. Only when it is more efficient than customers, CDN can become a valuable and sustainable industry.

Center and edge two-pronged approach to build CDN based on edge computing

Zong said that edge computing is a logical trend. For any cloud company, edge computing must be laid out in advance. Only by preparing for technical reserves, platform reserves and operational capabilities can we grasp the future.

When cloud computing companies look for application scenarios for edge computing, they will find that CDN is a typical application scenario. So why should CDN build on edge computing? Zong said: "Kingsoft Cloud builds CDN on the edge computing, not only can expand its resource pool, but also effectively improve the ability of CDN nodes to meet the elastic scalability of resources. For example, Akamai, the world's leading CDN service company, already has a lot of practice in the integration of edge computing and CDN. All of its CDN nodes are based on flexible and scalable container edge computing. All CDN companies will continue to take this step in the next year or two to expand and share their abilities."

So, will the rise of edge computing be inconsistent with previous central computing? Will it eventually lead to loss of competitiveness of the core business? Zong said that in the Internet industry, people always like to talk about who can replace it. But in fact, the two are more complementary to each other, to play their respective advantages to promote the sound development of the entire Internet. For example, in the process of CDN development, the demand of the central IDC has never been less, but it has been increasing. Because the overall demand is expanding. The relationship between cloud computing and edge computing is synergistic. The Internet industry is still a fast-changing market. For example, the original center computing requires a hashrate of 100. Now the hashrate demand is expanded to 1000, with 200 at the edge and 800 at the center. In this way, the hashrate demand of the center is expanded by 8 times.

So Zong Jie believes that there is no conflict between edge computing and so-called central cloud computing. They must be complementary to each other and will together meet the rapid development needs of the Internet.

Kingsoft Cloud actively faces the trend of CDN pipeline

In his speech, Zong mentioned that the second phase of the CDN architecture change is characterized by edge storage and CDN pipeline. Edge storage is very prominent in the CDN development architecture. In response to the explosive growth of customer content, storage clusters are implemented within the CDN, even across nodes in the same region. In response to the explosive growth of customer content, storage clusters are implemented within the CDN, even across nodes in the same area. In this way, the customer's edge storage capacity can reach PB, ensuring that the content is cached at the edge to maximum extent, thus achieving access to the nearest.

At the same time, CDN also presents a pipeline feature. For this trend, Kingsoft Cloud's approach is to face and embrace it. Zong shared how to embrace it in two major ways:

First, does the pipeline need to be infinitely large? This is impossible to achieve in reality. First, does the pipeline need to be infinitely large? This is impossible to achieve in reality. The first step is to make the pipeline more flexible, and then provide more flexible scheduling capabilities, greatly improving the utilization of the pipeline. The amount of water in different places, how it flows, where to build a pool, how big it is, and how to build a large pool in a short time, these are the key issues to make the pipeline more flexible.

The second is transparency. “The so-called transparency is to let customers know how much water can flow in your pipeline, whether the water quality is good or not, and whether the water flow rate can meet needs. This makes it easier and more reliable to use. It also improves resource efficiency. "

The future development of value-added services: AI-based CDN

Kingsoft Cloud started early in video cloud and AI empowerment fields, so its advantages are obvious. Zong said: "Kingsoft Cloud's development of AI empowerment is mainly to provide better quality and more stable service at a lower cost. This is what we have been working on all the time."

Kingsoft Cloud always see the pain points of customers and bring real benefits, such as its products KSHD and KIE. Zong said: "Kingsoft Cloud provides a one-stop service such as lightweight SDK, which is convenient for users. Under the premise of ensuring the same sense, it can reduce costs, improve service quality, and simplify the process. In just one step, users can experience the value of AI empowerment."

Zong stressed that the value of AI is not only a value-added service on pipeline, but the development direction of future Internet services. Akamai's value-added services already account for a large percentage of revenue, including security services. Zong said that security services will play a large demand in the enterprise market. However, at this stage, the investment in security has not yielded significant benefits. Kingsoft Cloud is currently focusing on AI. Among them, AI empowering video cloud is an important direction. It is the key to directly helping users save costs. Therefore, AI is the value-added service with the largest investment and the fastest return for Kingsoft Cloud at present. Overall, as for Kingsoft Cloud, the upwards development of CDN is AI, and downward is edge computing.