July 2019
Intermediate to advanced
502 pages
14h
English
Content preview from Hands-On Microservices with Kubernetes
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
Start your free trial
Vertical pod autoscaling
The vertical pod autoscaler is currently (Kubernetes 1.15) in its Beta stages. It takes on a different task related to autoscaling – fine-tuning your CPU and memory requests. Consider a pod that doesn't really do much and needs 100 MiB of memory, but it currently requests 500 MiB. First of all, it's a net waste of 400 MiB of memory that is always allocated to the pod and is never used. However, the impact can be much greater. Because the pod is chunkier, it can prevent other pods from getting scheduled alongside it.
The vertical autoscaler addresses this problem by monitoring the actual CPU and memory usage of pods and adjusting their requests automatically. It also requires that you install the metrics server.