Basic concepts and background of CDN

CDN Basic concepts and background

For CDN, you can refer to the following terms: Content distribute network, which can be directly translated into a content delivery network or written into content
Delivery Network: Content Delivery network. Obviously, CDN completes the task of transferring content from the source station to the user end. Of course, we do not need to explain what is "content delivery" or "content delivery, what is the value of CDN in the delivery or delivery process? Why does CDN need to be delivered instead of directly delivered through the Internet? It is necessary to give a short talk on this topic, which is helpful for understanding the working principles and key technologies of CDN in this book.

The Internet is a broad Internet, which consists of two layers: a network layer represented by TCP/IP (also a narrow Internet concept ); the other layer is the application layer represented by www. At present, there is a common misunderstanding that involves mixing the Internet and the World Wide Web. Recognizing the essence of the Internet and the differences between the Internet and the World Wide Web are the key and foundation for understanding the Internet Economy and the basis for understanding CDN.

The narrow Internet (Internet) with TCP/IP as the core is actually the lower layer of the Internet in the broad sense. It is the network base, and more generally the TCP/IP network. The main function of this layer is to transmit data packets of various information at extremely low costs through computer interconnection, which is commonly known as "pipelines ", all information and content are transmitted in this pipeline. The design concept of the Internet is: the network is neutral and uncontrolled, and no one has the right to decide. The network is not related to applications. Its task is to better transmit data packets end to end. This design concept has never been shaken since the beginning of the Internet. Any attempt to optimize the Internet for a certain type of content is not very effective. Therefore, we can think that the Internet will not try to optimize the transmission of any content.

The Application Layer represented by WWW is the upper layer of the broad-sense Internet. This layer includes many types of traffic and applications, such as email, software, online video, games, e-commerce, and mobile applications.
Providers and service providers are all visible and tangible applications provided by these users. They enrich and facilitate people's lives and constitute what we call Internet business and information economy.

Let's take a railway example to explain the difference and relationship between the two: the Internet is a railway track and signal system, and the World Wide Web is one of the trains running on the railway. In addition to the high-speed train World Wide Web, there are also slow trains, commuting trains, freight trains, and professional maintenance trains. On the Internet, the World Wide Web is huge and important, but it is not the only one. Applications that do not use WWW run on the Internet, and the Internet is far greater than anything that runs on it.

Now, let's see if there is a problem between the upper and lower layers of the network layer and the application layer. On the network layer, four parts of the Internet-based railway network can cause congestion.

 

(1) the "first kilometer" refers to the first exit for the world wide web traffic to be transmitted to users and the bandwidth provided by the link connecting website servers to the Internet. This bandwidth determines the access speed and concurrent traffic that a website can provide for users. The more users a website provides, the higher the egress bandwidth requirement. When the amount of data requested by a user exceeds the egress bandwidth, congestion will occur at the egress. When the business is busy, the more users access the service, the more congested the website will lose users when they need to provide services to users.

(2) "last mile" refers to the last access link, that is, the user access bandwidth, that is, the traffic sent from the World Wide Web to users. The average access bandwidth of users is one of the decisive factors affecting the development of upper-layer applications on the Internet. In the early stages of the development of the Internet, users access the Internet through dial-up or ISDN, and the network access speed is very low. Therefore, the Internet content is dominated by text with very small bandwidth usage, telnet and BBs were all mainstream applications at that time. After the emergence of the World Wide Web, human-computer interaction is more convenient and friendly multimedia content began to spread on the Internet, access bandwidth has become the main bottleneck restricting users to use the Internet. Starting from 2001, telecom operators began to develop broadband access services such as ADSL. With the increasing bandwidth and diversified access methods (fiber-optic entry, Wi-Fi, 3G, etc ), in recent years, the "last mile" problem has been greatly improved, especially in the past two years, China Telecom and others have vigorously carried out broadband acceleration services by means of fiber access, the bottleneck of "last mile" has been basically solved.

(3) Peer-to-Peer interconnection gateway. "Peer-to-peer interconnection" refers to the interconnection between different basic operators. Generally, there are only two or three interconnection points between two operators, you can imagine how much traffic is generated at these two or three points. When a website server is deployed in the IDC of carrier A, users of carrier B must access the website through the interconnection points between A and B. From the perspective of the Internet architecture, the interconnection bandwidth between different networks is relatively small for the network traffic of any operator, and the convergence ratio is very high, therefore, it is usually a congestion point in Internet transmission.

(4) Long-distance backbone transmission. The first is the long-distance transmission latency problem. The distance between the website server and the user is very far away from the IDC, backbone network, the user's man, and the user's access network, therefore, it is inevitable that a long transmission latency will affect the user experience. This problem cannot be solved by the Internet itself. The second is the backbone network congestion problem. Because most of the traffic on the Internet needs to be transmitted through the backbone network, the bearer capacity of the backbone network must be synchronized with the application of the Internet, however, the two are not in sync. When the upgrade and expansion of the backbone network lags behind the development of applications on the internet, it will gradually make the bearer Capacity of Large backbone networks a bottleneck affecting Internet performance.

In the application layer, SP are constantly optimizing the service experience, and the most noteworthy is the service response time. The service response time consists of the server response time and network latency. Factors that affect server response time include protocol processing time, program performance optimization, and content read speed, the network latency is the total latency caused by data packets being forwarded by vrouters and vswitches during network transmission. There is an "8-Second Law" in the Internet field. when a user visits a website, if the waiting time for the webpage to open exceeds 8 seconds, more than 30% of the users will give up waiting. According to a recent kissmetrics survey, a website cannot be opened 10 seconds later, and 40% of users will jump out of the page. Most mobile users are willing to wait for 6 ~ 10 seconds; 1 second delay will cause the conversion rate to decrease by 7%. If an e-commerce website generates 0.1 million yuan a day, a one-second page delay will cause it to lose 2.5 million yuan a year.

The emergence of CDN is closely related to a series of problems analyzed above. If these problems are not mitigated, the entire Internet will be a completely different scene from today. 1995, professor at MIT, Internet inventor tim
Berners-Lee predicted that Internet users would become the biggest obstacle to Internet applications in the future due to the congestion problem that Internet users were used to at that time. As a result, he challenged his colleagues to invent a brand new and fundamental solution to push Internet content. His proposal created CDN, an Internet basic service widely accepted today.

 

This article is excerpted from "CDN technical details"

By Lei Xiaohua, Sun Ying, Wang Feng, Chen Xiaoyi

Book details: http://blog.csdn.net/broadview2006/article/details/7690184