Disaggregated RIBs for BGP Resiliency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current BGP routing protocols face challenges in maintaining resiliency during updates or crashes, leading to potential network disruptions and resource-intensive state replication, which can cause churn and impact network stability.
Innovation Solution
Implementing a separate data store to isolate BGP peers from routing information churn, allowing for updates and crash recovery without disrupting the network, by receiving and storing incoming routing information in one data structure and generating state information for distribution in another, decoupled from the route selection process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If BGP peers maintain full state replication for resiliency, then network reliability improves, but resource consumption and network churn increase
Solution Approach 1:
The patent segments the BGP routing information base into multiple independent components (Adj-RIB-In, Loc-RIB, Adj-RIB-Out) that can be independently managed and replicated. This allows selective state replication of only critical components during failover scenarios, rather than replicating entire routing tables, thereby reducing resource consumption while maintaining reliability.
Solution Approach 2:
The patent implements selective copying of BGP state information between peers. Instead of full state replication, only essential routing state data is copied and maintained at backup peers, enabling crash recovery without the overhead of complete state synchronization across the network.
2Adaptability or versatility
If BGP protocol components are updated, then protocol adaptability improves, but network stability deteriorates due to exposure of changes to peers
Solution Approach 1:
The patent segments the BGP protocol implementation into distinct functional modules corresponding to different RIB components. This modular architecture enables independent updating of specific protocol components without requiring simultaneous updates across the entire protocol stack or notification to all peers, thus maintaining network stability during upgrades.
Solution Approach 2:
The patent extracts the routing information base structures from the BGP protocol implementation details. This separation allows the RIB data structures to be updated, modified, or upgraded independently from the BGP protocol state machine and peer communication logic, enabling protocol adaptability without exposing changes to network peers.
3Measurement precision
If BGP maintains detailed routing state information, then routing precision improves, but device complexity increases
Solution Approach 1:
The patent divides the routing information base into distinct segmented structures (Adj-RIB-In for received routes, Loc-RIB for locally selected routes, Adj-RIB-Out for advertised routes). Each segment maintains only the specific state information necessary for its function, preserving routing precision while avoiding the complexity of a monolithic data structure that would need to track all possible state variations in a single component.
Data Source
AI summary
The challenge of isolating a protocol peer(s) from routing information churn caused by to a peering protocol (e.g., due to updating the peering protocol, due to a bug in the peering protocol, due to a crash in the peering protocol, etc.) is solved by using a separate data store to isolate the protocol peer(s) from the peering protocol. The separate data store may: (a) receive, from at least one of the outside peering devices, incoming routing information; (b) store the incoming routing information received in a first storage system; (c) provide a copy of at least some of the stored incoming routing information received to a second storage system used by a process for selecting routes using the routing information, the process generating state information to be distributed (e.g., one or more routes with BGP properties (which may include one or more selected routes) and a set of outside peers to distribute them to) and storing the state information in the second storage system; (d) receive a copy of the state information generated and stored by the process; (e) store the received copy of the state information in the first storage system; and (f) provide at least some information from the copy of the state information to at least one outside peer device in accordance with a route advertisement process, regardless of a state or status of the process for selecting routes.