Cloud Interdatacenter Network Performance dataset

This dataset contains network performance data related to both AWS and Azure inter-datacenter networks, collected between March and November 2015. The collection process required more than 790 hours of syntethic traffic generation. We considered the 12 combinations of four regions selected for each provider (North Virginia, Ireland, Sao Paulo, and Singapore). Experiments have been run between VMs of the same size (M or XL). Repeated, 5-minute-long experiments have been performed in the same conditions, equally spaced in 24-hour intervals. According to the presence of multiple availability zones in each region for Amazon, we run around 8.6K and 880 5-minute-long experiments for Amazon and Azure, respectively. Beside performance measures, path information about each scenario has also been collected to complement the view on the performance. The directory tree identifies the specific experiment. In more details: provider └── VMsize and L4Protocol └──Region └──availability zone (only por AWS) └──file.json each json file contains experiments, reporting a number of network performance metrics (throughput, latency, jitter, packet loss), considering both synthetic and detailed results. json files are self-described. When refering to our data set, please cite the following reference: Valerio Persico, Alessio Botta, Pietro Marchetta, Antonio Montieri, Antonio Pescapè: On the performance of the wide-area networks interconnecting public-cloud datacenters around the globe. Computer Networks 112: 67-83 (2017)

本数据集涵盖2015年3月至11月期间采集的亚马逊网络服务（AWS）与微软Azure跨数据中心广域网的网络性能相关数据。数据采集过程累计耗时超过790小时，期间生成了模拟流量。本次实验选取了两家云服务商各4个区域，共计12种跨区域组合方案，涉及区域包括美国北弗吉尼亚、爱尔兰、圣保罗及新加坡。所有实验均在规格一致的虚拟机（M型或XL型）之间开展。实验以5分钟为周期重复进行，且在24小时间隔内均匀排布，全程保持实验条件一致。鉴于亚马逊网络服务各区域内设有多个可用区，我们分别为AWS和Azure开展了约8.6千次与880次5分钟时长的实验。除网络性能指标外，本数据集还采集了各实验场景的路径信息，以全面完善对网络性能的观测视角。数据集通过目录树结构标识具体实验，详细层级如下：服务商 └── 虚拟机规格与四层协议（L4Protocol） └── 区域 └── 可用区（仅适用于AWS） └── 实验数据文件.json。每个JSON文件均包含多组实验数据，涵盖吞吐量、延迟、抖动、丢包率等多项网络性能指标，同时包含模拟流量与细粒度实验结果。JSON文件自带说明信息。在引用本数据集时，请援引以下文献：Valerio Persico, Alessio Botta, Pietro Marchetta, Antonio Montieri, Antonio Pescapè: 《全球公有云数据中心互联广域网的性能研究》, 《计算机网络》112卷: 67-83 (2017)

搜集汇总

背景与挑战

背景概述

该数据集记录了2015年AWS和Azure跨数据中心网络的性能指标，包含8600次亚马逊实验和880次微软实验，涵盖四大洲区域。数据以结构化JSON格式存储，包含吞吐量、延迟等关键网络参数，适用于云计算网络性能分析研究。