Research On Campus Network Redundant Traffic Measuring Technology

With the development and innovation of information technology, the Internet, as the media of information dissemination, has gradually permeated into people’s daily life and production. On a specific communication link, retransmission of the same or similar resource will certainly bring about a large number of redundant traffic consuming a lot of bandwidth, which increase communication costs and reduce bandwidth utilization.Based on the real campus network, this paper does research mainly on two aspects:method of measuring technology with granularity of bytes chunk and analysis of the measuring results in campus redundant traffic. The main research contents are as follows:(1) Aiming at the requirement of packet samples’integrality, a high-speed packet capturing software of Linux platform was developed on the basis of the technology of PF_RING, which can wire-speed match the bidirectional traffic so as to achieve the integrity and effectiveness of the packet samples.(2) This paper puts forward a redundant traffic measurement method based on uniform sampling and dynamic tracking technology. The uniform sampling technology is realized as following steps:firstly, dividing continuous bytes chunks of packet payloads with specific size and computing corresponding hash fingerprint sequence with Rabin-Fingerprint method; secondly, continuously sliding a fixed-sized window on the sequence and selecting the characteristic fingerprint within the corresponding window by the established rules, which can ensure the corresponding bytes chunks evenly distributed on the packet payloads. The dynamic tracking technology is realized by the combination of buffer’s "First-In, First-Out" aging mechanism and iteratively updating index pointers of characteristic fingerprints.(3) According to the research on practically collected packets samples, this paper verifies the effectiveness of redundant traffic measurement method based on the uniform sampling and dynamic tracking technology, and confirms the existence of campus redundant traffic. From a macro perspective, this paper finds that redundant traffic shows self-similarity characteristic in time sequence, production intensive distribution within the area of source ports and asymmetric distribution in the business traffic of C/S model. From the microscopic point of view, this paper finds that redundant traffic has selectivity of the size of redundant bytes chunk. The fully redundant bytes chunks account for low proportion while contribute more identified redundant traffic. The characteristics of redundant traffic distribution provide theoretic reference b ases for improving current method of redundant traffic measurement.