A device and method for multiple load balancing processing

Abstract

The invention discloses multi-load equalization device and method. The multi-load equalization device comprises a configuration information base, a dynamic detection device, a resource calculating device, a task allocation device and a disaster backup device, wherein the configuration information base is used for storing static information of each server in a server cluster, a task allocation rule and algorithm information; the dynamic detection device is used for storing real-time information of each server in the server cluster; the resource calculating device is used for calculating a logic resource value or a logic matching value of a task according to a calculation request; the task allocation device is used for receiving tasks from a service node; and the disaster backup device is used for backup and disaster recovery of the load equalization device. With the adoption of the technical scheme of the invention, centralized control on load equalization and communication among the server clusters is realized, the resource utilization rate becomes average, and the longitudinal and transverse expansibility and robustness of a whole application system are also improved.

Description

technical field [0001] The invention relates to the technical field of computer data processing, in particular to a multiple load balancing device and method. Background technique [0002] With the continuous development of information technology, application systems based on server clusters are used more and more widely. For large enterprises, the processing timeliness and cost control of cluster systems are a major aspect of improving enterprise benefits. Therefore, the load of server clusters Balancing technology has become a key means for large-scale cluster systems to improve operational efficiency and allocate resources reasonably. At present, the load balancing of server clusters is mainly realized through direct communication between a service node and multiple proxy nodes. This method has certain defects in application systems that require multiple load balancing, which are mainly reflected in the following aspects: [0003] 1. The allocation of lower-level logical...

AI Technical Summary

Problems solved by technology

[0003] 1. The allocation of lower-level logical resources is not even enough: Since the current load balancing technology is for direct communication between a service node and multiple proxy nodes, the resources owned by each service node cannot be shared with each other. If multiple loads are performed from top to bottom In a balanced application system, if the balancing logic of each layer is different, it is easy to cause unbalanced load of the lower agent nodes

For example, the logical processing of an application system is divided into three layers, the first layer is file scanning, the second layer is initialization parameters, and the third layer is file processing, then in the load balancing from data scanning to initialization parameters (first Layer load balancing), the balancing logic is file number balancing, in the load balancing of initialization parameters and file processing (second layer load balancing), the balancing logic is file size balancing, and there is no necessary relationship between the number of files and the file size , it is easy to have the load imbalance of the second layer

[0004] 2. The vertical scalability of load balancing is low: since load balancing is currently implemented on service nodes and proxy nodes, if you want to increase load balancing vertically (that is, expand the level of load balancing), you must set up new service nodes and proxy nodes. And realize the communication, logical judgment, disaster recovery and other mechanisms between nodes, the scalability is relatively low

[0005] 3. The operation process of disaster recovery is relatively complicated and there are hidden dangers: when a service node fails, the current recovery method is to first enable the backup service node to take over the tasks of the failed service node, and then restore all agent nodes under the failed service node to Hang on to the backup service node, and then judge the progress of the tasks that have been run on the faulty service node and the proxy node. If the faulty service node can no longer provide the current task progress of the node and its proxy node, it is impossible to judge whether the existing job has been completed , there are hidden dangers

Images