.NET Architecture & System Design Interview Question #187

What is horizontal autoscaling?

Deployment, Cloud Operations, SRE & Senior System Design Scenarios Senior Advanced

Quick Interview Answer

Horizontal autoscaling automatically changes the number of application instances according to workload or another signal.

Detailed Explanation

Horizontal autoscaling automatically changes the number of application instances according to workload or another signal.

CPU can be useful, but queue depth, request rate, latency, or custom business metrics may better represent demand. Scaling rules should account for startup time, downstream capacity, minimum and maximum limits, and scale-in safety.

Code Example

Load increases
   |
Metric exceeds target
   |
3 instances -> 6 instances

Load falls
   |
safe scale-in