Cluster Level or Node Level Auto-Scaling

As we are getting ready for production level setup in Rancher Kubernetes.
So we want node auto-scaling in and out when resources gets insufficient.
As per the Rancher documentation there was not written anywhere directly that Rancher supports Node Level scaling but Rancher promises to handle production ready setup and it is one of the main part of that. So, what are the indirect and different solution/ways, which are currently available, to achieve this problem?