Loading educational lab…
LET’S BUILD WHAT’S NEXT
Let’s connect
Loading educational lab…
A simplified Horizontal Pod Autoscaler model. Desired replicas come from CPU demand versus a target. Latency numbers are educational, not a production performance prediction.
3 ready · 0 pending · 3 desired Demand exceeds Ready capacity. Scaling lags by startup time and cluster slots.
3Ready
0Pending
3Desired
128%CPU vs request
148 msEducational latency
94Capacity RPS
Desired ≈ ceil(ready × utilization / target), then clamped to min, max, and cluster slots. Latency is a teaching curve, not a measured p95.