Auto-scaling Tools
Automatically adding or removing capacity in response to load.
No autoscaling exists. Replica counts change only by operator action (`dh scale app`).
What auto-scaling actually automates
Auto-scaling watches a metric (CPU%, request queue depth, response latency) against thresholds and adds or removes running instances automatically -- scale out when load rises, scale back in when it drops, so you pay for capacity roughly matched to actual demand rather than provisioning for peak load permanently. Kubernetes' Horizontal Pod Autoscaler and cloud providers' auto-scaling groups are the common implementations; the hard parts are avoiding thrashing (scaling up and down rapidly on noisy metrics) and handling the ramp-up delay (a new instance takes real time to boot and become ready, during which load is still high).
Automatically adding or removing capacity in response to load. No autoscaling exists. Replica counts change only by operator action (`dh scale app`).