DCNetBench

DCNetBench是由中国科学院开发的首个面向不同用户、底层技术和不同基准测试需求可扩展的数据中心网络基准测试工具。该数据集通过构建模拟系统，运行实际应用以捕捉真实的网络流量模式，并通过对这些模式的分析和分类来建模和重放这些跟踪。DCNetBench提供了一个自动化的基准测试框架，支持这一流程，并在数据中心网络基准测试中展示了其可扩展性、有效性和多样性。该数据集主要用于解决数据中心网络性能评估和优化的问题，特别是在网络流量难以建模且被互联网服务巨头视为机密的情况下。

DCNetBench, developed by the Chinese Academy of Sciences, is the first scalable data center network benchmarking tool tailored for diverse users, underlying technologies and various benchmarking requirements. This benchmark dataset constructs simulated systems, runs real-world applications to capture authentic network traffic patterns, and models and replays traffic traces through analysis and classification of these patterns. DCNetBench provides an automated benchmarking framework that supports this workflow, and demonstrates its scalability, effectiveness and versatility in data center network benchmarking. This dataset is primarily designed to address the challenges of data center network performance evaluation and optimization, especially in scenarios where network traffic is difficult to model and is regarded as confidential by major Internet service giants.