Predictive scaling provisions capacity before demand arrives based on historical patterns.
How it works:
Analyze historical traffic patterns
Identify recurring cycles (daily peak at am)
Schedule capacity increases before predicted demand
Methods:
- Schedule-based: Scale up every weekday at :am
- ML-based: AWS predictive scaling uses machine learning
Best practice:
Combine predictive (handles known patterns) with reactive (handles unexpected spikes). Predictive for baseline, reactive for variance.