Considering an hybrid scenario based on on-premises systems and VM/services on Google Cloud, is it possible to shape the traffic over the Google Dedicated Interconnect link (e.g. define different priorities, different bandwidth, etc.)?
Are additional SW/HW needed at the end-points? Are there some technical constraints that need to be considered?
Thanks, Luca
It is not possible for now to manage traffic shaping directly on GCP side. Even though a Google Dedicated Interconnect link requires you to setup a Cloud Router (virtual router) on GCP’s end, this router’s sole function is to exchange BGP routes between GCP and your on-premises router for network discovery reasons.
That being said, an alternative to this scenario could be to setup a VM (with any vRouter software solution such as PFsense or OPNsense) behind the Cloud Router which would enforce the required QoS Policies.
You can also request this feature to be added to the GCP Cloud Router: https://cloud.google.com/support/docs/issue-trackers
It is possible to specify bandwidth per VLAN, see https://cloud.google.com/interconnect/docs/how-to/dedicated/creating-vlan-attachments. These are not hard policers, so you may see traffic bursting above configured limits, and there are no priorities per VLAN. But you can implement multiple VLANs with different bandwidths to approximate traffic shaping. There are no special sw/hw needed to implement, just a switch/router for on-prem.