RSVP-TE Graceful Restart via Extended HELLO TTL
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing RSVP-TE graceful restart mechanism is unable to establish a HELLO session between non-directly connected network elements, which prevents the preservation and recovery of label switched paths (LSPs) during control plane failures or restarts, especially when Fast Reroute (FRR) is in effect and the Point of Local Repair (PLR) or Merge Point (MP) nodes are not directly adjacent.
Innovation Solution
Establishing a HELLO session with a remote node that is not immediately adjacent by exchanging HELLO messages with a TTL value greater than one, allowing the messages to be forwarded through intermediate nodes and establishing a connection via IP forwarding or a tunnel, enabling the exchange of Restart_Cap objects to preserve and recover LSP states during control plane failures or restarts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If RSVP HELLO messages are exchanged only between directly connected neighboring nodes, then the HELLO session establishment follows the standard RSVP protocol specification, but the graceful restart cannot be achieved when PLR or MP nodes are not directly adjacent during FRR conditions
Solution Approach 1:
The patent introduces an intermediary mechanism by allowing HELLO messages to be forwarded through intermediate nodes via IP forwarding or tunnels. The TTL field acts as a mediator to control message propagation scope, enabling HELLO sessions between non-directly-connected nodes while preventing unlimited propagation. This resolves the contradiction by extending HELLO session establishment beyond direct neighbors without breaking protocol standards.
Solution Approach 2:
The patent changes the TTL parameter of HELLO messages from the conventional value of 1 to values greater than 1 (up to 255). This parameter change allows HELLO messages to traverse multiple hops through intermediate nodes, enabling graceful restart capability between non-adjacent PLR and MP nodes during FRR conditions, while the TTL constraint prevents unbounded message propagation.
2Ease of operation
If HELLO message TTL value is set to one, then the message is processed only by directly connected nodes as per RFC specification, but the message cannot reach remote nodes for graceful restart
Solution Approach 1:
The patent modifies the TTL parameter from 1 to values greater than 1, enabling HELLO messages to reach remote nodes through intermediate forwarding. The message forwarding remains simple because intermediate nodes use standard IP forwarding mechanisms without requiring special processing logic, thus maintaining ease of operation while achieving remote node detection.
Solution Approach 2:
The patent enables HELLO messages to self-forward through the network using standard IP forwarding mechanisms. Intermediate nodes automatically forward messages based on their routing tables without requiring special RSVP-TE processing, making the system self-service and maintaining operational simplicity while extending reach to remote nodes.
3Reliability
If HELLO messages are forwarded through intermediate nodes with TTL > 1, then graceful restart can be achieved between non-adjacent nodes, but the message propagation scope increases beyond direct neighbors
Solution Approach 1:
The patent uses TTL parameter to control message propagation scope. By setting TTL to specific values (e.g., 255), the system achieves graceful restart across FRR tunnels while limiting propagation to a controlled number of hops. This simple parameter-based control avoids complex propagation management mechanisms.
Solution Approach 2:
The TTL field acts as an intermediary control mechanism that automatically limits message propagation scope. Intermediate nodes use the TTL value to determine whether to forward messages, providing automatic propagation control without requiring complex device logic or additional control mechanisms.
4Reliability
If RSVP GR procedures are applied during FRR, then LSP state can be preserved during control plane failure, but no HELLO session exists between PLR and MP nodes when not directly connected
Solution Approach 1:
The patent introduces an intermediary HELLO session mechanism that bridges non-adjacent PLR and MP nodes through intermediate forwarding. This enables LSP state preservation during control plane failures by establishing the necessary HELLO session, while maintaining compatibility with existing RSVP GR procedures and RFC specifications.
Solution Approach 2:
The patent makes the HELLO message mechanism universal by enabling it to function both for direct neighbor detection (traditional use) and for remote node detection across FRR tunnels (extended use). The same HELLO message format and processing logic serve both purposes, maintaining compatibility while extending functionality.
Data Source
AI summary
According to one embodiment, in response to at least one of a link failure and node failure of a protected label switched path (LSP), network traffic is switched onto a protection path according to a fast re-reroute (FRR) scheme. A HELLO session is established with the remote node that is not immediately adjacent to the network element, including exchanging one or more HELLO messages with the remote node, each HELLO message having a time-to-live (TTL) value of greater than one if IP forwarding is used. The TTL value of the HELLO message is set to one if the HELLO message is sent via tunnel. In response to a request to restart, a resource reservation protocol (RSVP) traffic engineering (TE) graceful restart (GR) procedure is performed using information obtained from the one or more HELLO messages exchanged with the remote node during the HELLO session.