Cluster Election via Connectivity Bit-Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing database cluster systems face challenges in optimally selecting a master node during communication failures, leading to sub-optimal cluster configurations and potential data inconsistencies due to simplistic node ID-based selection methods, which may result in smaller surviving clusters and inaccurate node eviction strategies.
Innovation Solution
Implementing a prioritized messaging protocol to distribute cluster topology representations as connectivity bit-vectors across nodes, allowing for the identification and election of a master node in the sub-cluster with the largest number of members, ensuring optimal cluster configuration and minimizing data inconsistencies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If node ID-based selection method is used to identify master node, then selection simplicity is improved, but cluster optimality deteriorates
Solution Approach 1:
The patent changes the selection parameter from simple node ID to a composite parameter based on sub-cluster size and node ID. This allows the system to maintain simplicity in the selection process while improving optimality by considering the size of the sub-cluster, ensuring that the largest surviving cluster is selected as the master cluster.
Solution Approach 2:
The patent performs preliminary classification of nodes into sub-clusters based on connectivity bit-vectors before the master node selection. This preliminary action allows the system to identify the largest sub-cluster in advance, ensuring that the optimal cluster is selected as the master cluster while maintaining a simple selection process.
2Device complexity
If connectivity information is not accurately tracked, then system complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent uses connectivity bit-vectors as a compact copy of the connectivity information between nodes. Each node maintains a bit-vector that represents its connectivity status with other nodes, allowing accurate tracking of node statuses without requiring complex distributed state management, thus maintaining low system complexity while achieving high measurement precision.
3Loss of time
If simplistic master node selection is used, then decision speed is improved, but data consistency deteriorates
Solution Approach 1:
The patent changes the selection parameter from simple node ID to a composite parameter that includes sub-cluster size. This allows the system to make a quick decision by comparing cluster sizes while ensuring data consistency by selecting the largest cluster as the master cluster, preventing split-brain scenarios and maintaining ACID properties.
Data Source
AI summary
Described is a system, method, and computer program product for performing elections in a database cluster, where cluster topology representations are distributed across the cluster nodes using a prioritized messaging protocol. The cluster topology representations may be implemented as connectivity bit-vector that provide information that identifies which nodes in the cluster are able to communicatively connect with other nodes.


