基于sketch的流量统计方法性能实验数据

本数据集主要面向可重构智能网卡中高速网络流量统计方法中策略优化，基于libpcap采集生成高速网络报文大小数据，参考Netflow流量统计模版，主要记录了报文五元组信息（源目的IP地址、原目的端口号、协议类型），以及报文特征，即报文长度，获得了100k条报文特征信息，用于基于sketch的流量统计算法中冷热流替换策略（阈值）的优化。This dataset is primarily intended for policy optimization in high-speed network traffic statistics methods for reconfigurable smart NICs. High-speed network packet size data is collected and generated using libpcap, with reference to the NetFlow traffic statistics template. It mainly records the five-tuple information of network packets, namely source and destination IP addresses, source and destination port numbers, and protocol type, as well as packet characteristics including packet length. A total of 100,000 packet feature records are obtained, which are utilized to optimize the hot/cold flow replacement strategy (threshold) in sketch-based traffic statistics algorithms.

搜集汇总

背景与挑战

背景概述

该数据集面向可重构智能网卡中高速网络流量统计方法的策略优化，基于libpcap采集生成高速网络报文数据，记录报文五元组信息和长度特征，共包含100k条报文特征信息，用于优化基于sketch的流量统计算法中的冷热流替换策略。数据集由中国人民解放军国防科技大学创建，数据量为53.33MB，涉及计算机科学技术和通信技术学科。