SCALING POLICIES:
1. Target Tracking (Recommended)
- Maintain CPU at 50%
- Simple configuration
2. Step Scaling
- Multiple thresholds
- Different actions per step
3. Scheduled Scaling
- Known traffic patterns
- Time-based
4. Predictive Scaling
- ML-based prediction
- Proactive scaling
COOLDOWN PERIODS:
- Default: 300 seconds
- Prevents rapid scale in/out
- Can customize per policy
Ngày tiếp theo: Disaster Recovery Strategies