Project Overview

This project demonstrates how machine learning can enhance the monitoring of complex systems, specifically data centers.

What We're Doing

We generate synthetic data that mimics real-world sensor readings from a data center, including:

• CPU utilization
• Temperature
• Network latency

Why It Matters

Data centers require continuous, proactive monitoring. Detecting anomalies—unusual patterns in sensor data—is critical because they can indicate:

• Impending hardware failures
• Overheating
• Network bottlenecks
• Security breaches

Early detection helps:

• Prevent downtime
• Optimize resource allocation
• Maintain operational stability

How It Works

We leverage NVIDIA RAPIDS AI tools:

• cuDF for GPU-accelerated data processing
• cuML for machine learning

Using K-Means Clustering, we:

1. Group similar sensor readings into clusters
2. Identify data points far from cluster centroids
3. Flag those points as potential anomalies

The Goal

To showcase an automated, scalable approach for detecting critical system deviations—an essential capability for maintaining robust and efficient data center operations.