oscillation
When an automatic control keeps overshooting, such as scaling out, then in, then out again, instead of settling.
In more detail
Oscillation happens when a scaling signal reacts to the change it causes: servers are added, load per server drops, servers are removed, load climbs. Cooldowns, stabilisation windows and a well-chosen signal stop the loop.
Related terms
- cooldown — A waiting period after a scaling action during which further scaling is held back, so the system can settle.
- scaling policy — A rule that changes a group’s desired capacity based on a metric, such as average processor use or requests per instance.
- Auto Scaling group — An EC2 feature that keeps a desired number of identical instances running, replacing failed ones and adding or removing them as load changes.
Taught in
- Elasticity and Scale — Cloud Infrastructure & Hyperscaler Architecture
- Auto Scaling Groups — Cloud Infrastructure & Hyperscaler Architecture