Data center network monitoring faces a trade-off between measurement granularity and performance overhead. Existing simulation tools for network research often provide fixed, limited monitoring capabilities, restricting researchers’ ability to evaluate new algorithms. We introduce QoS-Sim, a flexible and extensible network simulation platform built on NS3. QoS-Sim provides a comprehensive set of customizable monitoring metrics, enabling detailed analysis of network performance. A core component of our platform is QDynamics, a lightweight algorithm designed to capture fine-grained queue dynamics efficiently. QDynamics is designed with future hardware implementation in mind, minimizing storage and bandwidth overhead. Our evaluation shows that QoS-Sim provides powerful data monitoring capabilities for both congestion control and buffer management algorithms, and the QDynamics module within QoS-Sim accurately reports queue behavior while reducing data storage requirements by up to 73%.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

QoS-Sim: A Flexible Simulation Platform for Fine-Grained Data Center Network Monitoring

  • Yubing Li,
  • Qingyun Liu

摘要

Data center network monitoring faces a trade-off between measurement granularity and performance overhead. Existing simulation tools for network research often provide fixed, limited monitoring capabilities, restricting researchers’ ability to evaluate new algorithms. We introduce QoS-Sim, a flexible and extensible network simulation platform built on NS3. QoS-Sim provides a comprehensive set of customizable monitoring metrics, enabling detailed analysis of network performance. A core component of our platform is QDynamics, a lightweight algorithm designed to capture fine-grained queue dynamics efficiently. QDynamics is designed with future hardware implementation in mind, minimizing storage and bandwidth overhead. Our evaluation shows that QoS-Sim provides powerful data monitoring capabilities for both congestion control and buffer management algorithms, and the QDynamics module within QoS-Sim accurately reports queue behavior while reducing data storage requirements by up to 73%.