AWS IOPS Insight - Project Summary

Overview

AWS IOPS Insight is a modern web application that demonstrates intelligent time series forecasting for AWS RDS Storage IOPS (Input/Output Operations Per Second) autoscaling. In this hackathon project, we showcase how machine learning can be used to predict future storage performance patterns and optimize cloud infrastructure costs. In the 6 hour period, our team fetched the data from Redash and finetuned the AWS Chronos time-series model on a couple of instances (26 for now but plan to finetune on more for better accuracy), achieving great results as showcased in the demo. This enables more efficient usage of AWS resources, enabling greater customer satisfaction and business growth, all-while netting a positive sustainability impact.

Core Functionality

Time Series Data Upload: Users can upload CSV files containing historical IOPS data
Real-time Forecasting: AI-powered predictions of future IOPS usage patterns
Interactive Visualizations: Dynamic charts showing observed vs. predicted values
Prediction History: Track and download previous forecasting results

Slideshow: https://gamma.app/docs/AWS-IOPS-Insight-9e45b0eolaoof7y