| 摘要: |
| 基于四川省气象预警决策发布系统原有架构进行升级，在没有新增服务器和网络设备的情况下，通过Nginx服务实现Web应用层的负载均衡，减轻网站应对高并发访问的压力。新增两个Redis数据缓存服务分别减轻数据库读取压力和提高网关读取海量任务数据的效率。新增消息队列组件实现发布任务至网关模块间的负载均衡和复杂模块的解耦，确保异常中断后的任务恢复。此次升级从软件层面以最小成本实现系统的性能优化，避免因为高并发阻塞导致的业务中断。升级后压力测试和实际应用情况表明，系统运行稳定，实现了负载均衡和性能优化的目标。 |
| 关键词: 负载均衡；Redis；消息队列；预警决策 |
| DOI： |
| 投稿时间：2021-03-09修订日期：2021-08-05 |
| 基金项目: |
|
| Load Balancing and Performance Optimization of Sichuan Meteorological Early Warning and Decision-making Publishing System |
| dan jia |
| (Sichuan Meteorological Service Centre) |
| Abstract: |
| Based on the upgrading of Sichuan meteorological early warning and decision-making publishing system, the load balance of web application layer is realized by Nginx service without additional server and network equipment, which reduce the pressure of website to deal with high concurrent access. Two new Redis services, one to reduce the database reading pressure, one to improve the efficiency of gateways reading massive tasks. Two new message queues implement load balancing between publishing tasks and gateway, decoupling from complex module, and to ensure task recovery after exceptional interrupt. This upgrade optimizes the performance of the system at the software level at minimum cost and avoids the interruption caused by high concurrent access. The pressure test and actual running proved that the system runs stably, the goals of load balancing realization and performance optimization is achieved. |
| Key words: load balancing; Redis; message queue; early warning and decision-making |
