节点文献
内容传播节能机制及随机Petri网评价
Energy Conservation Mechanism for Content Dissemination and Evaluation Analysis Based on Stochastic Petri Nets
【作者】 王峰;
【导师】 王高才;
【作者基本信息】 广西大学 , 计算机应用技术, 2013, 硕士
【摘要】 随着网络负载的急剧加大,网络内容传播过程变得更为复杂,传统上解决该问题的方法是增加网络设备数量和加大网络设备功率,这就使得内容传播过程中的能耗问题日益突出,所以,综合考虑网络内容传播中能耗和性能平衡是一个非常重要的研究主题。内容分发网络是一种主流的网络内容传播方式,能为用户提供较好的网络访问速度和稳定性,但其复杂和繁多的网络设备使得能耗相对较大,本文主要研究网络内容传播过程中加入节能机制的能耗分析和性能评价。本文针对内容分发网络的内容传播过程不同部分会出现不同时间尺度的空余时间,提出了多级睡眠-唤醒机制,在低负载或者其他一些情况下,将设备的一些部件转变成睡眠状态以到达节能效果,并且使得节能最大化与睡眠状态相关数据损失最小化。对提出的多级睡眠-唤醒机制运用随机Petri网进行能耗分析,在随机Petri网建模的基础上应用模型化简和评价方法,通过分析随机Petri网模型各状态稳定概率得到系统平均能耗,并使用软件SPNP对加入节能机制前后的内容传播过程能耗进行对比分析。最后通过加入节能机制前后的内容分发网络SPN模型的性能分析,判断节能机制是否对内容传播过程中的性能产生影响。仿真实验结果表明多级睡眠-唤醒机制能很好地降低内容传播过程中的平均能耗,不过会对内容传播过程中的响应时问产生一定的影响。本文针对内容分发网络的内容传播过程中,内容传播路由、分发内容过程中的链路负载情况时刻都在发生变化,通过结合考虑性能指标因素定义转变链路速率机制,提出了基于最大延迟阈值的速率调节机制的网络节能策略,该机制使得链路速率保持最长时间的最低速率。对提出的基于最大延迟阈值的速率调节机制运用随机Petri网进行能耗分析,通过加入节能机制前后的内容分发网络SPN模型进行性能评价,判断该机制是否对内容传播过程中的性能产生影响。使用软件SPNP进行仿真分析,仿真实验结果表明该机制使得CDN系统节能最大化而数据延迟没有明显的增加,但该节能机制对吞吐量有一定的影响。本文提出的多级睡眠-唤醒机制和基于最大延迟阈值的速率调节机制都不仅仅只是考虑节能问题,而且将满足内容传播过程中的性能指标综合到节能问题一起考虑。因此,提出的节能机制有很好的实际意义。提出的节能机制对于内容传播方法具有普遍意义,这两个机制也能够应用于研究其他内容传播方法的节能问题上。
【Abstract】 With the rapid increase of network load, making the process of content dissemination becomes more complex, traditionally the solution to this problem is to increase the number of network equipment and network equipment power, which makes the problem of energy consumption in the process of content dissemination increasingly prominent, So considering the balance between network content transmission energy consumption and performance is a very important research topic. Content Delivery Network is a mainstream kind of network content dissemination method, it can provides users with better network access speed and stability, but the complex network equipment has consumed large energy consumption, This thesis mainly studies energy consumption analysis and performance evaluation of the network content dissemination progress after join the energy saving mechanism.In this thesis, in views of the process of content dissemination in Content Delivery Network, different parts will be time free in different time scales, the author proposed the multistage sleep-wake up mechanism, which at low load or some other cases, some equipment parts will transfer into the sleep mode in order to reach energy saving effect, and reach the goal of maximize saving energy and minimize loss data that associated with sleep state. Using Stochastic Petri Net to analysis the multistage sleep-wake up mechanism’s energy consumption, applying model reduction and evaluation method based on Stochastic Petri Net modeling, by analyzing stable probability of Stochastic Petri Net model to get system average energy consumption, and use the software SPNP to analyze the process of content dissemination’s energy consumption, further more, make a comparison between before and after join the energy-saving mechanism in the process of content dissemination. Finally, judging whether energy saving mechanism makes influence on the content dissemination process performance, by analysis before and after join the energy-saving mechanism into the Content Delivery Network SPN model’s performance. Simulation experimental results show that multistage sleep-wake up mechanism can reduce the average energy consumption in the process of content dissemination, but it will produce certain effect on response time in the communication process.In views of the process of content dissemination in Content Delivery Network, the link load changes all the time in the process of content dissemination routing and content distributing, by considering performance index factors defined link rate change mechanism, the author proposed the Max-Delay-Threshold ALR mechanism for network energy consumption, to make link rate keep the longest minimum rate. Using Stochastic Petri Nets to analysis the Max-Delay-Threshold ALR mechanism’s energy consumption, judging whether energy saving mechanism makes influence on the content dissemination process performance, by analysis before and after join the energy-saving mechanism into the Content Delivery Network SPN model’s performance. Using software SPNP for simulation and analysis, simulation experimental results show this mechanism make the CDN system energy efficiency maximization but increase the content data delay a little, but it has certain impact on throughput.In this thesis, the multistage sleep-wake up mechanism and the Max-Delay-Threshold ALR mechanism are both not just considering energy conservation issue, but also consider performance index in the process of content dissemination together to the energy conservation problem. Therefore these energy saving mechanisms both have great practical significance for real situation and universal significance for content dissemination method. These two mechanisms can also be applied to the research on other content dissemination method’s energy consumption.
【Key words】 Content dissemination; Stochastic Petri Nets; ContentDistribution Network; Energy conservation strategy; Energyefficiency analysis;