BGP Speaker Controlled Failover Mechanism
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for deploying Border Gateway Protocol (BGP) in high availability scenarios, such as stateful switchover and graceful restart, are inefficient and do not scale well as the volume of routes increases, leading to prolonged downtime during failover and potential missed topology changes.
Innovation Solution
A mechanism is introduced to pause BGP message transport, transfer the BGP state and transport state from an active to a standby speaker, and then resume operations, allowing for a controlled failover that minimizes downtime and synchronizes the standby speaker quickly, enabling it to take over without disrupting network processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If bulk data transfer is used to synchronize routing tables from Active to Standby BGP speaker, then high availability is achieved, but the approach does not scale as route volume increases and causes prolonged downtime
Solution Approach 1:
The patent segments the bulk routing table synchronization into incremental updates. Instead of transferring the entire routing table at once, the system divides route information into individual route entries that are transferred separately as they are generated or modified, enabling continuous synchronization without prolonged downtime
Solution Approach 2:
The patent implements preliminary action by having the Standby BGP speaker synchronize with the Active speaker before failover occurs. Route information is pre-transferred and cached in advance, so when failover happens, the Standby speaker already possesses the necessary routing data to immediately assume the Active role without waiting for data transfer
2Reliability
If all data structures are transferred and flattened for inter-process communication, then stateful switchover is enabled, but device complexity increases and conversion overhead grows
Solution Approach 1:
The patent extracts the essential routing information from complex data structures, transferring only the necessary route entries rather than entire data structures. This selective extraction eliminates the need for flattening and conversion of complex pointer-based structures, reducing complexity while maintaining switchover capability
Solution Approach 2:
The patent uses simplified copying mechanisms for route information transfer. Instead of complex data structure conversion and flattening, the system creates simplified copies of route entries that can be directly transferred and reconstructed, reducing conversion overhead and structural complexity
3Quantity of substance
If incremental route updates are transmitted as changes occur, then data transfer volume is reduced, but synchronization overhead increases due to frequent RIB updates
Solution Approach 1:
The patent implements periodic action by batching incremental route updates and transferring them at scheduled intervals or when a threshold is reached, rather than transferring every single change immediately. This periodic synchronization reduces the frequency of update operations while maintaining data freshness, lowering synchronization overhead
Solution Approach 2:
The patent applies partial action by transferring only the necessary subset of route information that has changed, rather than synchronizing the entire routing table. This selective transfer of partial data reduces both data transfer volume and synchronization overhead compared to full table updates
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method and apparatus for recovering from a controlled failover of a BGP speaker is provided. A user sends, to a network element, a request to switch a designation of an active Border Gateway Protocol (BGP) speaker of the network element from a first BGP speaker to a second BGP speaker. After receiving the request, the network element pauses operation of a transport for BGP. Routing data, which describes a state of a first routing information base (RIB) maintained by the active BGP speaker, is transferred from the first BGP speaker to the second BGP speaker. Thereafter, the first BGP speaker may instruct the second BGP speaker to become the active BGP speaker. After the second BGP speaker becomes the active BGP speaker, the second BGP speaker resumes operation of the BGP transport.