I found AWS's autoscaling offering be better than GCP's, in that you could use custom CloudWatch metrics as a basis for scaling logic rather than just the stand

I found AWS's autoscaling offering be better than GCP's, in that you could use custom CloudWatch metrics as a basis for scaling logic rather than just the standard metrics (CPU usage, latency, etc) GCP does offer custom metrics, but I couldn't figure out how to tie them to the LB scaling logic.