可用带宽监控模型的设计与实现

发布时间：2018-05-30 01:19

本文选题：云计算 + 服务质量　；参考：《电子科技大学》2014年硕士论文

【摘要】：随着计算机科学和网络技术的不断发展,近年来出现了以云计算为代表的新兴技术,云计算技术在提供大量高性能服务的同时,也对承载着这些服务的传输网络提出了新的要求。对高性能网络监控系统的需求也应运而生。本文针对上述需求,以云计算最关注的性能指标之一,带宽为研究课题,深入理解网络监控,重点研究了带宽测量技术的原理和方法、网络监控技术的种类和特点以及适用于网络带宽的预测分析模型等,主要研究内容分为以下几个部分。首先,对传统的带宽测量工具的原理和采用的技术进行了分析。基于探测包时延的带宽测量技术开启了带宽测量的先河,但该技术忽略了网络背景流量的突发性,且多次采样求平均平滑误差的效果不明显,导致该类算法精确度较低。为了提高测量算法的精确度,引入了基于统计学原理的探测技术,该方法具有接入带宽门槛低、受背景流量影响小的特点,能够提高测量精度。其次,云计算环境下的应用大多对服务质量敏感且对可用带宽的要求较高,基于统计的测量算法虽然测量精度高,但无法根据实时的可用带宽来调节自身的测量力度,在可用带宽不足时有可能会发生探测流量干扰正常业务流的情况。因此,提出了自适应探测速率的可用带宽测量算法,该算法能够根据实时可用带宽来调节探测流量,还能在可用带宽低于阈值时发出预警。第三,研究了网络监控和预测技术。为了实时地监控网络中各段链路的可用带宽,并且预测可用带宽的变化趋势,设计出了网络带宽监控预测模型。该模型根据监控需求对监控功能进行了模块的划分,测量模块借助带宽测量算法对可用带宽进行监控,分析模块通过数据处理对测量结果进行监控,而反馈模块则根据监控结果和预测分析技术对将来一段时间内可用带宽的变化趋势进行了判断。最后,为了实现网络带宽监控预测模型,对云操作系统中的云资源监控模型进行了扩展,将模型中的各个模块重新封装,增加了可用带宽测量和预测的功能,提升了云平台运行的可靠性。为了验证该模型,在服务器中搭建实验环境,模拟了多个实验,分别验证了模型的监控准确性、自适应性和预测分析功能。实验结果表明,该模型在可用带宽监控和可用带宽预测方面都具有良好的性能。
[Abstract]:With the continuous development of computer science and network technology, cloud computing technology, represented by cloud computing, has emerged in recent years. Cloud computing technology provides a large number of high-performance services at the same time. It also puts forward new requirements for the transmission network carrying these services. The demand for high performance network monitoring system also arises at the historic moment. In order to meet the above requirements, this paper focuses on the principle and method of bandwidth measurement technology, which is one of the most concerned performance indexes of cloud computing, and takes bandwidth as the research topic, deeply understanding network monitoring, and focusing on the principle and method of bandwidth measurement technology. The types and characteristics of network monitoring technology and the prediction and analysis model suitable for network bandwidth are mainly studied in the following parts. Firstly, the principle and technology of traditional bandwidth measurement tools are analyzed. The bandwidth measurement technology based on the detection packet delay opens the first step of the bandwidth measurement, but this technique ignores the sudden occurrence of the network background flow, and the effect of multiple sampling to average smoothing error is not obvious, which leads to the low accuracy of this kind of algorithm. In order to improve the accuracy of the measurement algorithm, the detection technology based on the principle of statistics is introduced. This method has the characteristics of low threshold of access bandwidth and small influence of background flow, which can improve the accuracy of measurement. Secondly, most applications in cloud computing environment are sensitive to the quality of service and require higher available bandwidth. Although the measurement algorithm based on statistics has high measurement accuracy, it can not adjust its measurement intensity according to the real-time available bandwidth. Detection traffic may interfere with normal traffic when the available bandwidth is insufficient. Therefore, an adaptive available bandwidth measurement algorithm for detection rate is proposed. The algorithm can adjust the detection flow according to the real-time available bandwidth, and also can give an early warning when the available bandwidth is below the threshold. Thirdly, the technology of network monitoring and prediction is studied. In order to monitor the available bandwidth of each segment of the network in real time and predict the trend of the available bandwidth, a network bandwidth monitoring and forecasting model is designed. The model divides the monitoring function into modules according to the monitoring requirements. The measurement module monitors the available bandwidth with the help of bandwidth measurement algorithm, and the analysis module monitors the measurement results through data processing. The feedback module judges the trend of available bandwidth in the future according to the monitoring results and predictive analysis techniques. Finally, in order to realize the network bandwidth monitoring and prediction model, the cloud resource monitoring model in the cloud operating system is extended, each module in the model is re-encapsulated, and the function of available bandwidth measurement and prediction is added. Improved the reliability of cloud platform operation. In order to verify the model, an experimental environment was built in the server, and several experiments were simulated, respectively, to verify the monitoring accuracy, adaptability and predictive analysis function of the model. Experimental results show that the model has good performance in both available bandwidth monitoring and available bandwidth prediction.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.06

【参考文献】