Artificial intelligence-based network traffic analysis and automatic optimization technology

Area of Science:

• Computer Science
• Network Engineering
• Data Center Management

Background:

• Current data center network operation and maintenance (O & M) relies on device status checks, which fail to capture real business data transmission.
• This limitation prevents a comprehensive understanding of overall business running conditions and service bearing capacity.

Purpose of the Study:

• To introduce a novel method for enhancing data center network O & M by analyzing TCP packet transmission.
• To enable O & M engineers to gain a comprehensive perception of service bearing status within data centers.

Main Methods:

• Implementation of Encapsulated Remote Switch Port Analyzer (ERSPAN) technology to mirror TCP packets.
• Application of stream matching rules in the packet forwarding path.
• Utilization of a network O & M AI collector for in-depth TCP packet analysis, traffic statistics, path recapture, delay computation, and application identification.

Main Results:

• Development of an end-to-end visualized correlation model linking networks and services.
• Enabled comprehensive perception of service bearing status for O & M engineers.
• Provided a foundation for data center optimization and proactive network risk warnings.

Conclusions:

• The proposed ERSPAN-based method significantly improves the visibility of network and service status in data centers.
• This approach facilitates a tighter coupling between network performance and service delivery.
• It offers robust technical support for data center optimization and early detection of network risks.