Large-scale cluster node performance monitoring system

Abstract

The invention relates to a large-scale cluster node performance monitoring system. The system is characterized by comprising a monitoring management platform, an agent node cluster and an agent client. According to the performance monitoring system provided by the invention, the monitoring management platform is in charge of distributing and scheduling monitoring tasks of all members in the agent node cluster, and monitoring that configuration files are pushed by the monitoring management platform to the members in the agent node cluster in a unified mode; and redundancy examination and data synchronization are performed on all nodes in the agent node cluster, when agent nodes fail, the monitoring management platform, according to loads of residual nodes, distributes monitoring tasks which are originally executed by the failed agent nodes to other normal nodes, the process does not affect or interrupt monitoring, and thus normal operation of the monitoring system is effectively guaranteed.

Description

technical field [0001] The invention relates to a performance monitoring system of large-scale cluster nodes, in particular to a performance monitoring system which is safe, highly available and convenient for large-scale dynamic expansion of large-scale cluster nodes. Background technique [0002] The current data center administrators generally know and supervise the operation of the server in a timely manner through the monitoring system, and quickly solve the problem when the server is abnormal, so as to ensure the safety, stability and reliability of the server. Monitoring information also provides important reference data for business operation analysis and improvement. [0003] The usual framework of the current mainstream monitoring system is a two-tier architecture model of monitoring server and proxy client or a three-tier architecture of monitoring server, proxy node and proxy client. Regardless of whether there are agent nodes or not, these mainstream monitoring...

AI Technical Summary

Problems solved by technology

Regardless of whether there are agent nodes or not, these mainstream monitoring systems mostly use dual-machine backup technology nodes for server redundancy, so the expansion is limited. Secondly, the multiple agent nodes of these monitoring systems monitor their respective targets and do not communicate with each other. If a proxy node is damaged, all its monitoring services will be interrupted

Therefore, the existing monitoring system is difficult to meet the environment with huge monitoring resources, especially the environment that cannot adapt to the mature application of cloud computing technology.

Images