Server Power Consumption Data for Failure Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Predicting failures in remote servers is challenging due to the distance of vendor resources, leading to potential delays in repair and loss of computing resources, as existing methods rely heavily on health alerts that may not accurately indicate upcoming component failures.
Innovation Solution
A method that analyzes power consumption data trends and compares them with a server failure model derived from historical data to predict failures, using a deep neural network for pattern recognition and sending alerts for impending failures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If health alerts are used to predict server failures, then failure prediction is provided, but the prediction accuracy is insufficient and may not accurately indicate upcoming component failures
Solution Approach 1:
The patent changes the monitoring parameter from health alerts to power consumption data. By analyzing trends in power consumption data and comparing them with a server failure model, the system achieves more accurate failure prediction. This parameter change enables detection of subtle changes in server behavior that precede failures, improving both reliability and measurement precision.
2Ease of operation
If vendor resources are distant from remote servers, then centralized management is achieved, but repair time increases and computing resources are lost
Solution Approach 1:
The system performs preliminary actions by continuously analyzing power consumption data to predict failures before they occur. When a potential failure is detected, the system can proactively replace components or take preventive measures, eliminating the need for lengthy repair processes. This transforms the reactive repair model into a proactive maintenance model, reducing downtime while maintaining centralized management.
3Reliability
If power consumption data is analyzed to predict failures, then early identification of potential failures is enabled, but additional data processing requirements are introduced
Solution Approach 1:
The patent introduces an intermediary server that collects, analyzes, and processes power consumption data from multiple remote servers. This intermediary system handles the complex data processing tasks, including comparing power consumption trends with failure models and generating predictions. By centralizing the analytical functionality, the complexity is managed systematically while enabling early failure identification across the server fleet.
Data Source
AI summary
A method for billing server utilization based on power includes receiving power consumption data of a remote server used by a customer, deriving a power-utilization correlation between power consumption of the remote server with utilization of the remote server, determining utilization of the remote server from the power consumption data and the power-utilization correlation, preparing a bill for the customer based on the determined utilization of the remote server, and sending the bill to the customer.