Cluster Member Session Information Segmentation for Rapid Failover
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional router clusters face scalability issues due to the need for session information exchange among members, which increases communication loads and fails to achieve simultaneous increase in the number of router devices and reduction in failover time effectively.
Innovation Solution
A cluster system where each member executes session processing and packet forwarding without exchanging session information, with a failure recovery mechanism that allows a cluster member in charge of sub-processing to take over main processing using recorded session information when a failure occurs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If session information is exchanged among cluster members to enable failure recovery, then reliability is improved, but device complexity and communication load increase
Solution Approach 1:
The patent segments session information management by designating specific cluster members (those handling sub-processing) to maintain session information for packets they process, rather than requiring all members to exchange and maintain complete session information. This segmentation reduces overall communication overhead while ensuring failure recovery capability.
Solution Approach 2:
The patent introduces a traffic distribution filter as an intermediary mechanism that uses packet assignment information to determine which cluster member should handle specific packets. This intermediary enables targeted session information maintenance without requiring comprehensive information exchange among all cluster members.
2Productivity
If the number of router devices is increased to improve load distribution, then productivity is improved, but scalability is limited due to increased session information exchange requirements
Solution Approach 1:
By segmenting session information management responsibilities among cluster members based on packet assignment, the system can scale to more devices without proportionally increasing communication overhead. Each member only maintains session information relevant to its assigned packets.
Solution Approach 2:
The patent changes the parameter of session information exchange from a comprehensive all-to-all communication model to a selective model based on packet assignment information. This parameter change enables the system to scale more effectively by reducing communication complexity.
3Device complexity
If session information is maintained centrally to simplify management, then device complexity is reduced, but failover time increases
Solution Approach 1:
Cluster members perform preliminary actions by maintaining session information for packets they are assigned to process. This preliminary preparation ensures that when a failure occurs, the failover can occur rapidly without requiring time-consuming information exchange or reconstruction of session state.
Solution Approach 2:
Each cluster member serves itself by maintaining its own session information for assigned packets, eliminating the need for centralized session information management. This self-service approach reduces management complexity while enabling rapid local failover decisions.
Data Source
AI summary
Distribution processing control unit determines a main processing range and a sub-processing range whose charges are taken by a cluster member to which the unit belongs. Among packets multicast to a cluster system, a filter hands over packets matching with the main processing range and the sub-processing range to a session processing unit. The session processing unit conducts session processing (including processing of updating session information stored in a holding unit). Thereafter, when the packet matches with the main processing range, the session processing unit hands the packet over to a packet forwarding unit and when the packet matches with the sub-processing range, abandons the packet. When a cluster member in charge of main processing of a packet whose sub-processing is taken charge of by the cluster member in question develops a failure, a failure recovery unit changes the sub-processing range to the main processing range.