Distributed Node Group Identification in Split-Brain Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In a network split-brain situation, distributed computing environments face challenges in determining data ownership and maintaining service continuity and data integrity when a network is split into multiple groups, each assuming the others are down, leading to difficulties in rejoining as a single group.
Innovation Solution
A method is introduced to determine a major group by comparing the number of nodes in each split group and utilizing history information, such as node IDs and joining times, to dynamically adjust the quorum and resolve conflicts, ensuring service continuity and data integrity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional quorum algorithms are used to determine major group in split-brain situation, then the system can maintain fault tolerance, but it fails to accurately determine data ownership when groups have equal number of nodes
Solution Approach 1:
The system pre-assigns group identifiers to nodes before the split-brain situation occurs. When a network partition happens, these pre-assigned identifiers allow immediate determination of the major group without requiring complex real-time voting, thus maintaining both reliability and determination accuracy even when groups have equal node counts
Solution Approach 2:
The patent introduces group identifiers as an intermediary mechanism that mediates the major group determination process. Instead of nodes directly voting against each other, the pre-assigned identifiers serve as a neutral reference that objectively determines which group is the major group, resolving the ambiguity that occurs with traditional quorum algorithms
2Adaptability or versatility
If the network is split into multiple groups with equal number of nodes, then each group can independently operate, but it becomes impossible to determine which group owns the correct data
Solution Approach 1:
Group identifiers are assigned to nodes in advance, before any network partition occurs. This preliminary assignment ensures that even when the network splits into equally-sized groups, each node can immediately determine its group's identity and compare it with the pre-stored identifier to ascertain data ownership, preventing information loss about which group holds the correct data
Solution Approach 2:
Each node uses its own pre-assigned group identifier to independently determine whether its group is the major group. This self-service mechanism allows nodes to autonomously identify data ownership without requiring communication with other groups, maintaining group independence while preventing data ownership information loss
3Measurement precision
If history information is collected and stored for all nodes, then the system can accurately determine major group using history information, but the system complexity and storage requirements increase
Solution Approach 1:
The patent extracts only the essential history information needed for major group determination - specifically, the group identifier assigned to each node. By taking out only this critical piece of information and storing it in a simple key-value format, the system achieves accurate major group determination without the complexity of storing comprehensive node history
Solution Approach 2:
Each node stores only its own group identifier locally, rather than maintaining centralized history information for all nodes. This local quality approach distributes the storage burden and simplifies the system architecture while still enabling accurate major group determination when needed
Data Source
AI summary
The present invention relates to a method, system and computer readable recording medium for determining a major group in a network split-brain situation that occurs in a network-based distributed computing environment. According to an aspect of the present invention, there is provided a method of determining a major group among at least two groups created by splitting an initial group consisting of a plurality of nodes in a network-based distributed environment, the method comprising the steps of comparing the number of nodes included in a first group with the number of nodes included in a second group; acquiring history information of the nodes included in the first and second groups; and determining whether the first group or the second group is the major group using at least one of a result of comparing the number of nodes and the history information.