一种具有自学习功能的CPU流控模式探索

Insight of an Automatism Learning Mode for CPU Flow Control

介绍了分布式交换机系统环境下CPU流控模型与算法;通过综合考虑并引入CPU实时利用率、学习抑制因子、学习步长等,构造了一具有简单自学习能力的CPU流控算法,改进了传统算法完全依靠经验数据的弊端;同时考虑了算法在嵌入式设备上的可实现性和易用性。因此构造自学习算法模型,综合考虑在各种网络业务环境下如何实现CPU流控是算法的关键。在基于自学习和适应不同运行业务考虑的基础上,构造出了一个CPU流控的算法模型,并就算法在不同情况下的工作效能做了分析。

Presenting a model and arithmetic for CPUFC（CPU Flow Control）in distributed switching System.It mends the flaw of traditional arithmetic which just depends on experienced knowledge elements.While the new mode constructs one simple automatism learning arithmetic for CPUFC which considering following factors：learning restrain factor,learning step size,availability in embedded devices and implementation,etc.And the essential is how to construct this automatism learning model and build out the CPUFC arithmetic to adapt to various network services.The article provides a CPUFC mode based on automatism learning and analyzes it＇s working effect in different cases.