The invention discloses a data analysis system comprising a virus feature library, a regular inspection module, a virus feature matching module, a port auditing module, a traffic counting module, a network health judging module, a virus simulation module, an emergency channel module, a reduction module, and a data isolating and uploading module. The network traffic is monitored and audited in order to keep the network in good condition. Through analysis and memorization of unknown intrusion behaviors, the immune ability of the network is improved. After invasion, the hazard scope can be controlled effectively to ensure smooth network and normal service providing. The system has the ability to self-repair and -reduce, and can maintain stable operation of the network. Through use of the distributed technology and the load balancing technology, a large amount of network traffic data is jointly analyzed by multiple network traffic analysis engines, so the waiting time is reduced for users, and the overall performance of the system is improved. Moreover, the system has data protection ability, and the occurrence of information loss when a user is away from the computer is avoided.