Configuration data dynamic updating system and method based on application cluster

Abstract

The invention relates to a configuration data dynamic updating system and method based on an application cluster. The system and method has the advantages that a configuration file version number scheme is introduced; configuration data is classified and managed through a configuration file version number based on a cluster management server; the cluster management server informs application nodes of a version number change event when the configuration file version number changes; after learning the version number change event through monitoring, the application nodes acquire the latest version from the cluster management server, read configuration data corresponding to the latest version number, analyze the configuration data, assemble the analyzed data into local cache objects, and update local caches to complete dynamic updating of the configuration data; and only the configuration file version number is stored on the cluster management server, and the specific configuration data is stored through discs, so that limits from a memory of the cluster management server are removed, and the dynamic updating of massive data can be achieved.

Description

technical field [0001] The invention relates to the technical field of data updating, in particular to a system and method for dynamically updating configuration data based on application clusters. Background technique [0002] With the gradual popularization of distributed technology, there are more and more application systems based on distributed technology. In addition, Linux and virtual technology have become the current mainstream technology trend, and the application system is no longer limited to the deployment of minicomputers, but an application system cluster composed of multiple x86 blade machines or virtual machines. The application system cluster refers to the deployment of applications on multiple machines. Cluster deployment facilitates the horizontal expansion of the application system and linearly improves performance, but it also brings certain problems to O&M and management. For example, each node in the cluster is faced with configuration data ( Especia...

AI Technical Summary

Problems solved by technology

[0005] (1) Znode storage size is limited, the maximum can not exceed 1M, zookeeper client and server have restrictions

Although the maximum value can be increased by modifying the environment variable "jute.maxbuffer", if the ZNode is too large, reading and writing a certain ZNode will cause an uncertain delay, and if the ZNode is too large, the ZooKeeper server will be exhausted too quickly memory, so ZooKeeper is not suitable for storing large amounts of data;

[0006] (2) It is not suitable for transmitting a large amount of data, and the configuration data of the application system changes. Zookeeper needs to synchronize data with each node of the application system cluster in a short period of time (preferably at the millisecond level). When the number of nodes is large, the network bandwidth is insufficient, which will affect the application system. The heartbeat maintenance with zookeeper triggers the failure of the application and zookeeper session, which in turn will affect the processing of configuration data by the application (how to deal with the configuration data that was not synchronized before, and how to deal with the data in the local cache), after the session resumes, zookeeper will not The watcher function will be triggered again;

[0007] (3) Zookeeper is not suitable for storing structured data, because zookeeper's tree structure storage feature makes it difficult to express configuration data with more complex structures

Images