Database Cluster Master Node Selection via Failure Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing database cluster systems face challenges in optimally selecting a master node during communication failures, particularly when ties occur based on node count, leading to potential sub-optimal resource allocation and increased risk of cluster outages due to overloading.
Innovation Solution
The system uses resource statistics data to predict node failures by classifying anomalies, determining the probability of failure, and intelligently selecting a master node to minimize the likelihood of subsequent cluster failures, avoiding overloading nodes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a master node is selected based on simple node count criteria, then the selection process is simple and fast, but the selected node may be overloaded and prone to failure
Solution Approach 1:
The patent changes the selection parameters from simple node count to a comprehensive set of resource statistics including CPU utilization, memory usage, network bandwidth, and I/O operations. This transforms the master node selection from a simplistic criterion-based approach to a multi-parameter evaluation system that identifies nodes with lower failure probabilities.
Solution Approach 2:
The system performs preliminary analysis of resource statistics and anomaly detection before master node selection. By classifying anomalies and determining failure probabilities in advance, the system prepares failure prediction data that guides the master node selection process, avoiding the need for reactive selection after failures occur.
2Reliability
If resource statistics analysis is performed to predict node failures, then the reliability of master node selection improves, but the complexity of the selection process increases
Solution Approach 1:
The system implements self-service through automated anomaly detection and failure probability assessment. The cluster nodes autonomously monitor their own resource statistics, detect anomalies, and provide failure prediction information to the master node selection process without requiring external intervention or manual analysis.
Solution Approach 2:
The patent incorporates feedback mechanisms where resource statistics are continuously collected, analyzed for anomalies, and used to update failure probability assessments. This feedback loop enables the system to adapt to changing node conditions and refine master node selection based on real-time cluster health information.
Data Source
AI summary
Described is a system, method, and computer program product for performing elections in a database cluster, where system resource statistics information is used to predict a cluster node failure. Resource statistics data is classified and used to identify anomalies. The anomalies can be used to determine the probability of a cluster node failure and to then elect a new master node and/or surviving sub-cluster.