I have a Node.js app that provides a service which cannot be interrupted.

However the load to the app varies overtime and to save cost I would like the vm instance machine type to autoscale in function of the load (ie when over 80% CPU utilisation, scale up from 1 vCPU(3.75 GB memory, n1-standard-1) to 2vCPU(7.5 GB memory, n1-standard-2))

Is this possible?

PS: I have looked at using the container engine and kubernetes but due to how the app operates, the app cannot be replicated to multiple pods and continue working