Active-Active Service Router Cluster Resource Redistribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Determining which surviving service routers should assume responsibilities for providing services configured on a failed router in an active-active service router cluster is difficult and time-consuming, especially when all but two routers fail, leading to potential service disruptions and increased downtime.
Innovation Solution
An approach that allows for the quick and efficient selection of an active service router to assume responsibilities of a failed router by computing hash values based on rank information, enabling the redistribution of logical resources such as port IP addresses without explicit negotiations between routers, ensuring high availability and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If explicit negotiations are used between service routers to determine responsibility for failed router services, then the selection process can be thorough and coordinated, but the time required for service redistribution increases and service disruptions are prolonged
Solution Approach 1:
The patent applies preliminary action by pre-computing hash values for all possible failure scenarios before any actual failure occurs. Each service router pre-calculates which services it would assume if any other router fails, using a deterministic hash function based on router identifiers and service identifiers. This eliminates the need for time-consuming negotiations during actual failures, as routers already have predetermined responsibility assignments ready to execute immediately
Solution Approach 2:
The patent implements self-service by enabling each service router to independently determine its responsibility for failed router services without needing to negotiate with other routers. Each router autonomously computes whether it should assume a service by evaluating pre-computed hash values locally, eliminating inter-router communication overhead and accelerating the service redistribution process while maintaining consistent decisions across the cluster
2Productivity
If a thorough negotiation process is implemented to distribute services among surviving routers, then the distribution can be optimized, but the complexity of the failure response process increases and recovery time is extended
Solution Approach 1:
The patent applies parameter changes by transforming the complex multi-router negotiation problem into a simple deterministic computation using hash values. Instead of implementing complex negotiation protocols, the system changes the approach to using a hash function that takes router identifiers and service identifiers as inputs and produces a deterministic output that directly indicates responsibility assignment. This simplifies the failure response process while maintaining efficient service redistribution
Solution Approach 2:
The patent extracts the essential decision-making logic from the complex negotiation process and encapsulates it in a simple hash computation. By taking out the core responsibility-determination function and implementing it as a deterministic hash-based rule, the system eliminates the need for complex inter-router communication and coordination protocols, reducing process complexity while preserving redistribution efficiency
Data Source
AI summary
A method for a service router (“SR”) implemented in an active-active SR cluster to move logical resources without explicit negotiations between the service routers (“SRs”) in the cluster to provide high availability of services is disclosed. In an embodiment, the method comprises: determining, by an active SR of a plurality of SRs implemented in an active-active SR cluster, that a particular SR has failed; and obtaining, by the active SR, a plurality of hash values computed for the particular SR. The plurality of hash values comprises hash values for the active SR. The active SR determines a maximum hash value of the plurality of hash values, and if the maximum hash value corresponds to, and only to, the hash value determined for the active SR, then the active SR configures a port IP address of the particular SR on the active SR.


