Cluster flow control method and system

Abstract

The invention discloses a cluster flow control method and system. The cluster flow control method comprises the following steps: establishing a distributed cluster in a global network, counting the flow of each node in the global network, setting a node flow threshold value of each node, and regulating the actual passing request flow of each node through the node flow threshold value; periodicallycollecting the actual request traffic of the global network through a traffic monitoring program, dynamically monitoring the traffic threshold of each node, and predicting the actual request trafficof each node in the next period; and receiving the flow regulation and control information published by the flow monitoring program through a flow notification program, and transmitting the flow regulation and control information to each node of the distributed cluster. According to the cluster flow control method and system, the overall flow of the whole distributed cluster can be limited, the flow threshold of each node can be dynamically adjusted, and the performance bottleneck of a single node is prevented from influencing the pressure of the whole cluster network.

Description

technical field [0001] The invention relates to the field of network data communication, in particular to a method and system for controlling cluster traffic. Background technique [0002] At present, many cluster networks include several single nodes, which are used to transmit traffic data. However, with the in-depth development of big data, these cluster networks are faced with a large amount of instantaneous traffic. The traffic is limited. Due to the excessive request traffic, it is easy to cause a single node in the cluster network to be paralyzed, which in turn affects the normal operation of the entire cluster network. [0003] Therefore, there are two ways to solve this problem. The first way is to individually control the traffic threshold for each node in the cluster network. The traffic threshold of the entire cluster network is: the number of cluster nodes * the traffic threshold of a single node, but The disadvantage of this solution is: when the request in th...

AI Technical Summary

Problems solved by technology

[0002] At present, many cluster networks include several single nodes, which are used to transmit traffic data. However, with the in-depth development of big data, these cluster networks are faced with a large amount of instantaneous traffic. The traffic is limited. Due to the excessive request traffic, it is easy to cause a single node in the cluster network to be paralyzed, thereby affecting the normal operation of the entire cluster network

[0003] Therefore, there are two ways to solve this problem. The first way is to individually control the traffic threshold for each node in the cluster network. The traffic threshold of the entire cluster network is: the number of cluster nodes * the traffic threshold of a single node, but The disadvantages of this solution are: when the request in the cluster is skewed, the request volume of a single node is too large, but the request volume of the entire cluster is still small, and the flow control is not accurate

The second way is to count the cluster request volume by counting the services of a single node. By passing each request in the cluster network through this service, the service can determine the traffic threshold of the entire cluster network. The disadvantage of this solution is : The judgment of the service order of a single node is likely to cause the bottleneck point of the entire cluster network, and the reliability is difficult to guarantee

Images