Transfer learning of PaaS platform traffic monitoring via clustering
摘要
为了满足Web应用的快速部署
自动维护和自动扩容的需求
从而产生了PaaS平台。但随之而来的问题是如何实时监控PaaS的流量。为了能够实现流量的实现监控
研究人员提出了利用聚类算法来实现自动分类
但数据在传送很容易受到外界因素的影响
从而导致采集的流量是失真的
因此根据这样的数据来聚类分析后的结果是不准确的。针对此问题
以模糊C均值算法为基础
借鉴知识利用的思想
提出了一种具有迁移学习能力的聚类算法。并将其应用到PaaS平台的流量实现监控中
从而能够快速识别流量
从而能够从极大的保证系统的稳定安全的运行。
Abstract
The PaaS platform isstructured to realize the Web’srapid deploymentand to satisfythe need of Web’s maintain and dilatationautomatically. But
there is a urgent problem that how to monitor the flow of PaaS platformat any time.In order to be able to implement traffic monitoring related researchers using the clustering algorithm is presented to realize automatic classification
but the data in transmission is easily affected by external factors
which leads to acquisition of flow is distorted
so according to the data to the results of cluster analysis is not accurate.To solve this problems
a new clusteralgorithm
based on FCM algorithmand transfer learning thought
is introduced. This new cluster algorithm is used to PaaS platform and tomonitor the flow of PaaS platform at any time
so that can recognition flow quickly and can make platform run softly and stably.