基于计划驱动的图像数据处理系统云计算弹性伸缩设计  

Cloud Computing Elastic Scaling Design for Image Data Processing System Based on Plan-driven

作　　者：陈雪华 王刚 孙赫 郑美 CHEN Xuehua;WANG Gang;SUN He;ZHENG Mei(Beijing Institute of Remote Sensing Information,Beijing 100011,China)

机构地区：[1]北京市遥感信息研究所,北京100011

出　　处：《计算机与网络》2024年第3期194-199,共6页Computer & Network

摘　　要：图像数据处理系统资源构建在专有云上,针对系统任务负载非周期变化、任务计划强关联等特点,利用图像数据处理计划可预测业务负载的优点,设计了基于计划的任务量预测自定义指标,提出了基于计划驱动的弹性伸缩策略和系统设计。通过实验环境测试,实现了计算资源根据业务负载动态调整,进一步提高了资源利用率。同时增加了基于CPU负载的伸缩保底策略,提高了系统可靠性。Image data processing system resources are built on a private cloud.According to the characteristics of non-periodic changes of system task load and strong correlation of the task plans,and using the advantage of image data processing plans to predict business load,a user-defined index for task quantity prediction based on plans is designed,and an elastic scaling strategy and system design based on plan-driven is proposed.Through experimental environment testing,the dynamic adjustment of computing resources based on business load has been realized,and the resource utilization has been improved.At the same time,a scalable guaranteed strategy based on CPU load has been added to improve the system reliability.

关 键 词：云计算 图像数据处理 弹性伸缩 负载动态调整 

分 类 号：TP393.09[自动化与计算机技术—计算机应用技术]