AI 용어집
인공지능 완전 사전
DCIM (Data Center Infrastructure Management)
Software platform for unified monitoring and management of a data center's physical infrastructure, integrating power, cooling, and security to optimize energy efficiency.
PUE (Power Usage Effectiveness)
Performance indicator measuring the ratio between total energy consumed by a data center and energy actually used by IT equipment, serving as a benchmark for energy optimization.
Free Cooling
Passive cooling technique using external climatic conditions (air, water) to reduce energy consumption of air conditioning systems, often controlled by AI algorithms.
Liquid Cooling Immersion
Direct cooling method by immersing servers in a non-conductive liquid dielectric, eliminating ventilation needs and optimizing heat transfers.
Dynamic Workload Placement
AI strategy that intelligently distributes computing workloads in real-time to minimize energy consumption while respecting performance and latency constraints.
Predictive Thermal Modeling
Machine learning modeling that predicts thermal flows and hot spots in a data center to anticipate and optimize cooling needs.
Reinforcement Learning for Cooling
Application of reinforcement learning algorithms to autonomously control cooling systems by dynamically adapting to operational conditions.
Server Consolidation
AI optimization process that consolidates workloads on fewer physical servers to maximize utilization rates and reduce overall energy consumption.
Cold Aisle Containment
Physical isolation system for cold aisles to channel cooling air directly to server air intakes, improving overall thermal efficiency.
Energy-Aware Scheduling
Scheduling algorithm that incorporates energy costs and dynamic pricing in computational task planning to minimize the electricity bill.
Digital Twin for Data Center
Dynamic virtual replica of a data center using AI to simulate, analyze, and optimize energy operations before physical implementation.
Anomaly Detection in Power Consumption
AI-powered monitoring system identifying abnormal deviations in power consumption to detect equipment failures or energy waste.
Adaptive Voltage Scaling
Optimization technique that dynamically adjusts processor supply voltage based on workload to reduce energy consumption without compromising performance.
Heat Reuse System
Infrastructure capturing and utilizing residual heat from data centers for urban heating or other applications, optimized by thermal management algorithms.
AI-Powered Power Capping
Intelligent power consumption limitation mechanism per server, managed by AI to respect global data center constraints while maintaining critical performance.
Computational Fluid Dynamics (CFD) Simulation
Complex numerical modeling of airflow and heat distribution in a data center, accelerated by AI to optimize equipment layout and cooling strategies.
Green AI for Data Centers
Set of artificial intelligence techniques specifically designed to minimize the carbon footprint and energy consumption of computing infrastructures.
Real-time Energy Monitoring
System of continuous and instantaneous measurement of electricity consumption at all levels of the data center, powering energy optimization algorithms.
Workload-aware Cooling
Adaptive cooling strategy adjusting the air conditioning capacity based on current and anticipated workloads to eliminate energy oversizing.
Edge Computing Energy Optimization
AI-based optimization of task distribution between central data centers and edge infrastructures to minimize the overall energy consumption of the computing network.