Auto-splitting Cluster Subsystems for Link Failure Continuity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cluster management systems fail to effectively mitigate disruptions caused by communication link failures between sites within a cluster, leading to resource issues and partial cluster failures.
Innovation Solution
A method that identifies communication link issues and automatically splits the cluster into sub-clusters, with each site managing local workloads and storing data locally, and then merges the sub-clusters back into a single cluster when communication is restored.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the cluster operates as a single unified system, then resource sharing and coordination are optimized, but the system becomes vulnerable to communication link failures causing partial or complete cluster failure
Solution Approach 1:
The patent divides the cluster into multiple independent sub-clusters that can operate autonomously. Each sub-cluster contains its own set of sites that can function independently when communication links fail, preventing complete cluster failure and ensuring continuous operation of critical workloads.
Solution Approach 2:
The cluster structure dynamically adapts between unified and segmented states based on communication link status. When links are healthy, sites operate as a unified cluster for optimal resource sharing. When links fail, the system automatically segments into independent sub-clusters, and can re-merge when links are restored, providing dynamic response to changing conditions.
2Productivity
If the cluster automatically splits into sub-clusters during communication failures, then workload continuity is maintained, but system complexity and management overhead increase
Solution Approach 1:
The cluster implements self-service through automatic detection of communication link failures and autonomous splitting into sub-clusters without human intervention. The system monitors its own health status and automatically reconfigures itself, reducing management overhead while maintaining workload continuity during failures.
Solution Approach 2:
The system prepares for potential failures by pre-establishing the capability to split into sub-clusters. Sites are pre-configured with the knowledge and ability to operate independently, so when a communication failure occurs, the transition to sub-cluster mode is immediate and seamless, maintaining productivity without manual reconfiguration.
3Speed
If sites store data locally in sub-clusters, then data accessibility and processing speed are improved, but data consistency and synchronization become more difficult
Solution Approach 1:
Each sub-cluster is designed with local quality characteristics, storing and processing data independently within its own boundaries. This enables fast local data access and processing speed while maintaining data consistency through defined interfaces and protocols that ensure proper synchronization when sub-clusters re-merge or communicate with each other.
Data Source
AI summary
Methods, computer program products, and/or systems are provided that perform the following operations: identifying a first workload being processed by a first plurality of sites in a cluster; identifying, from the first plurality of sites: (i) a first site as a primary site for the first workload, and (ii) one or more secondary sites for the first workload; identifying a communication link issue between the first site and at least one of the one or more secondary sites; splitting the cluster into sub-clusters based, at least in part, on the communication link issue, wherein the first site is included in a first sub-cluster of the sub-clusters and the at least one of the one or more secondary sites is included in a sub-cluster of the sub-clusters that is different from the first sub-cluster; and instructing the first sub-cluster to locally process the first workload.