Explore MongoDB's groundbreaking predictive auto-scaling algorithm in this 29-minute conference talk by Jesse and Matthieu. Discover how MongoDB prevents overloads, reduces costs, and minimizes carbon footprint across half a million database servers on three public clouds. Delve into the future of cloud computing and machine learning, examining successful techniques and understanding the challenges and opportunities of implementing this revolutionary system. Learn about MongoDB's revenue model, Atlas scaling, and autoscaling processes. Investigate the components, data flow, and planning involved in the algorithm. Analyze seasonal patterns, MSTL, accuracy metrics, short-term forecasting, and CPU estimation. Gain insights into the results achieved and future plans for this innovative approach to database management.
Overview
Syllabus
Intro
MongoDB as a Service
MongoDB Revenue Model
Atlas Scaling
Autoscaling
Ideal future
Experiment
Components
Data Flow
Planning
Data Science
Seasonal Patterns
MSTL
Accuracy
Shortterm Forecast
CPU Estimator
Results
Future plans
Taught by
Data Council