I worked on a project with an e-commerce company a few months back. They use autoscaling in their Kubernetes cluster to account for high load during peak hours (generally noon until 7PM or so). It would go to 10-ish instances of 2 apps during this peak, and then during the non-peak times it'd drop back down to 2 instanches.

This is pretty significant, since the 2 different apps are relatively large JVM apps, each requiring ~16GiB of memory