GCP Compute Engine - is it possible to autosize resources?

Question

Recently I started managing a GCP project with 3 CentOS instances and 1 Windows Server. I'm trying to catch up with the documentation, but there's something I couldn't find anywhere.

One of those Linux instances is a host for an "almost-insane-resource-consuming" service. For the past 6 months the whole team (all developers, none from infrastructure per se) was increasing the resources manually only when the users started to complain about slow response times or failed uploads. My first response was to implement monitoring for all instances, which is now working flawlessly, but what I really wanted was to autosize CPU, disk and/or RAM. I've read about autoscaling, which isn't exactly what I need.

This instance is now running with 4 vCPU and 20GB RAM, which is currently considered "oversized" for GCP standards, since RAM usage is <20% for the past 7 days.

TL;DR: Is it possible to autosize (increase or decrease) resources on Compute Engine based on instance usage?

Answer 1

In theory, scaling instances vertically is easy. However, in the real world resizing an instance can result in failure. Disk resizing might fail, the zone might not have enough resources to launch an instance, etc.

Google does not offer feature to automatically resize vertically for Compute Engine.