Predictive Data Center Cooling for Server Heat Spikes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud-based storage systems face challenges in thermal management due to excessive heat, leading to increased failure rates of storage servers, as tenants have limited ability to proactively manage their devices within data centers and are often unaware of heat contributions from neighboring servers.
Innovation Solution
A predictive thermal model that uses server workload data, telemetry, and weather data to forecast temperature changes, enabling proactive management through operational changes such as altering I/O modes, adjusting cooling airflow, and reallocating workloads to prevent overheating.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If servers operate at high I/O operational modes to maximize productivity, then data processing speed improves, but temperature increases causing drives to fail prematurely
Solution Approach 1:
The predictive thermal model forecasts future temperature conditions before they occur, allowing the system to proactively adjust I/O operational modes. By predicting temperature trends based on historical data and current conditions, the system can prevent overheating before it causes drive failures, rather than reacting after the problem occurs.
Solution Approach 2:
The system dynamically adjusts I/O operational modes based on real-time temperature predictions and actual thermal conditions. The host can switch between different I/O modes (e.g., from high-performance to heat-responsive modes) to optimize the balance between productivity and thermal management, allowing flexible adaptation to changing thermal conditions.
2Temperature
If tenants proactively manage their devices to reduce heat, then thermal events are mitigated, but tenants have limited ability to control their devices within data centers
Solution Approach 1:
The system enables tenants to self-manage their devices by providing them with predictive thermal models and automated decision-making capabilities. The host can autonomously adjust I/O operational modes based on temperature predictions, allowing tenants to effectively manage their devices without requiring direct control over data center infrastructure or neighboring servers.
Solution Approach 2:
The predictive thermal model acts as an intermediary that bridges the gap between limited tenant control and effective thermal management. By using the model to forecast thermal conditions and automatically adjust operations, tenants can achieve heat management goals despite having restricted direct control over their devices within the data center environment.
3Reliability
If cooling capacity is increased to maintain lower temperatures, then drive reliability improves, but energy consumption increases
Solution Approach 1:
By predicting future temperature conditions before they occur, the system can proactively adjust I/O operational modes to prevent overheating. This allows for reduced cooling capacity during periods when thermal issues are anticipated and managed through operational changes, thereby reducing cooling energy consumption while maintaining drive reliability.
Solution Approach 2:
The system changes operational parameters (I/O modes) to manage thermal conditions rather than relying solely on increasing cooling capacity. By adjusting I/O operational characteristics in response to temperature predictions, the system can maintain drive reliability while minimizing the need for additional cooling energy consumption.
Data Source
AI summary
The present technology pertains to a predictive thermal model that can be used to intelligently manage thermal events in a data center. The predictive thermal model can be used to predict future temperatures of servers to take action before the server experiences higher than desired temperatures. The present technology also includes several innovative amelioration techniques that can help to keep servers cool when it is predicted that heat in their environment is about to increase. One such amelioration technique is a heat-responsive operation change for storage servers, or at least individual hosts within a storage server. For example, a host can be switched into a mode where it can batch read and write operations to limit the amount of seeking the host needs to perform, which produces less heat.