Flow management method and device for distributed CDN

Abstract

The invention discloses a method for managing distributed CDN flows. The method includes the following steps that states of all of nodes and the flows of all the nodes of a distributed CDN are monitored; when downtime happens on a first multiple node, a first up one level node covering the first multiple node is searched; whether a second multiple node covered by the first up one level node can bear the flows of the first multiple node or not is judged; if yes, the flows of the first multiple node are switched to the second multiple node; if no, a second up one level node on the upper layer of the first up one level node is searched and the method is returned to the C procedure. When the downtime happens on the node, the node with bearing capacity is automatically searched, rapid response to node faults in a computer room is achieved, and losses of the flows are reduced. The invention further discloses a device for managing the distributed CDN flows.

Description

technical field [0001] The invention relates to the technical field of the Internet, in particular to a traffic management method and device for a distributed CDN. Background technique [0002] In the field of Internet traffic management, virtual nodes and levels of virtual nodes are defined. Among them, a virtual node refers to a virtual concept including two or more actual nodes. These actual nodes have the following characteristics: they belong to the same operator, they are geographically close, they cover similar areas, and they contain service IPs of all physical nodes. Switching the traffic of the actual node to the virtual node can achieve the effect of polling the service IP of the physical node in proportion. [0003] Virtual nodes can have different levels. For virtual nodes of the same level, the coverage area and covered physical nodes do not overlap, and the coverage area of ​​a high-level virtual node and the actual nodes contained are the union of multiple...

AI Technical Summary

Problems solved by technology

[0010] (1) If the node is down when the traffic has just been switched but is not stable yet, it may lead to inaccurate traffic estimation;

[0011] (2) If there is a sudden increase or decrease in traffic due to changes in user behavior, the actual traffic may not match the estimated traffic

[0012] The above two problems cannot be solved by traditional traffic management methods for the time being, and manual intervention is required

The manual adjustment of the static allocation strategy is based on uncertain factors such as the feeling and experience of the personnel for allocation management, which not only has low management efficiency but also low accuracy

Images