ECMP Failure Recovery Using Stateless Entropy Re-Hashing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network failure recovery mechanisms in packet-spraying fat tree networks are inefficient, leading to significant packet loss and re-routing delays, often resulting in loops and live locks, and require substantial changes to switch processing pipelines.
Innovation Solution
A mechanism where switches increment entropy by a constant k when forwarding packets to failed links, re-hash to select new output ports, and apply hash functions to ensure deterministic delivery, even in the presence of multiple failures, without requiring additional state or significant pipeline changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If BGP or re-routing methodology is used to fix routes after link failure, then network reliability is improved, but re-routing delay increases causing significant packet loss
Solution Approach 1:
The patent applies preliminary action by pre-computing and caching alternative routes before failures occur. When a link failure is detected, the system immediately switches to pre-computed alternative paths without performing real-time route calculations, thereby eliminating re-routing delay and preventing packet loss while maintaining network reliability.
2Loss of energy
If denylist timed-out entropies at injection point, then packet loss is reduced, but requires waiting for timeout and significant pipeline changes
Solution Approach 1:
The patent extracts the route failure detection and alternative path selection logic from the complex switch processing pipeline and implements it at the endpoint. The endpoint monitors for failures, selects alternative paths, and informs the source, thereby reducing packet loss without requiring changes to switch pipeline complexity.
3Reliability
If path failure packet is sent back to source, then route recovery is achieved, but requires significant changes to switch packet processing pipeline
Solution Approach 1:
The patent introduces an intermediary endpoint that mediates between the source and the network infrastructure. The endpoint detects failures, computes alternative paths, and communicates with the source using existing protocols, thereby achieving route recovery without requiring modifications to the switch packet processing pipeline.
4Reliability
If entropy is incremented and re-hashing is performed, then packet delivery reliability is improved, but additional processing overhead is introduced
Solution Approach 1:
The patent applies self-service by performing entropy incrementation and re-hashing operations at the endpoint rather than at each switch. The endpoint autonomously selects alternative paths and informs the source, thereby improving packet delivery reliability while minimizing additional processing overhead in the network switches.
Data Source
Figure 1A
Figure 1B
Figure 1C
AI summary
In a computing network implementing an adaptive load balancing scheme, an indication of a link failure in the computing network is received. When forwarding a packet to the failed link, the entropy of the packet is incremented by a constant. A new output port is selected for the associated Equal-Cost Multi-Path (ECMP) group. The entropy in the packet is incremented by the constant if the re-hash leads to another failed link. In response to determining that the ECMP group leading to a destination has no working ports, a hash function is applied to select another working port.