High-efficiency filtering method for data packets

Abstract

The invention discloses a high-efficiency filtering method for data packets. The method comprises the steps as follows: firstly, preprocessing all filtering rules to obtain a filtering rule retrieval table and a mapping relation table of the field type filtering rules with the same number as CPU (central processing unit) cores; secondly, obtaining the filtered field information of the data packets to be processed by the CPU cores and determining the filtering rules corresponding to the filtered field information through a dextroposition operation and a binary search algorithm; thirdly, processing the filtered field information in the data packets according to the corresponding filtering rules so as to finish filtering of the data packets. According to the method, the filtering rules can be quickly and accurately matched, so that the filtering speed is greatly increased, the data throughput performance is greatly improved, and the data packet filtering efficiency is improved.

Description

technical field [0001] The invention relates to the technical field of digital information transmission, in particular to a digital information processing method. Background technique [0002] With the rapid development of Internet technology, the amount of data information on the Internet is not only increasing, but also more and more diverse. While network users require stable network data flow, they also hope to adopt specific filtering strategies for some specific data information. Therefore, one of the technologies that can realize these requirements is the data packet filtering method: in the huge data information, only the required data packets are selected, and the unnecessary or even threatening data packets are filtered. Usually the method of data packet filtering is to formulate the required filtering rules according to the field content of the data packet, such as source / destination IP address, data packet length, IP protocol, port number and application layer i...

AI Technical Summary

Problems solved by technology

However, the existing data packet filtering method based on multiple domains consumes too much resources due to matching filtering rules and cannot be processed in parallel without locks. Performance requirements under diversity and huge data traffic, especially on backbone network nodes, this performance requirement is more obvious

People try to use various methods to solve this problem. Among them, the efficient and representative ones are the HASH table packet filtering method, the RFC recursive flow classification packet filtering method and the simple binary tree search filtering method, but these methods have limitations. Disadvantages: Although the HASH table packet filtering method is easy to implement, has small storage requirements, and supports range matching and dynamic update, its efficiency is greatly affected by the distribution of rules and the number of rules. When there is a conflict, the search time is relatively long; Although the RFC recursive flow classification packet filtering method is fast, has good scalability, and is easy to process in parallel, but because the performance of the recursive flow classification packet filtering method is related to the structure of the classifier, this method lacks generality and needs to be based on different The best performance can only be achieved by fine-tuning the classifier, and this method consumes a lot of memory resources; the simple binary tree search and filter method is convenient for hardware implementation, the rule dimension has good scalability, supports range matching and dynamic update, but the search time increases with the number of rules Explosive growth, not suitable for high-speed network applications in the case of multi-dimensional

Images