How to Place and Size Cloud Interconnect VLAN Attachments

GCP Study Hub
November 18, 2026

A VLAN attachment connects a Cloud Interconnect connection to a Cloud Router in a region. Each attachment rides on one connection, but one Dedicated Interconnect connection can carry more than one VLAN attachment, and each attachment can connect to a Cloud Router in a different region on the same continent. A connection therefore isn't limited to the region nearest its colocation facility. That flexibility is behind many cost, latency and capacity decisions on the Professional Cloud Network Engineer exam, because a new or resized attachment can often do what would otherwise take a new circuit.

Reaching another region from an existing connection

Because attachments aren't tied to the nearest region, an existing connection can reach a new region on the same continent through a new VLAN attachment in that region. Traffic rides the same private physical link and Google's backbone to the workloads there, so the reach of a connection grows without another physical connection behind it.

Placing attachments where the workloads run

Egress through Cloud Interconnect is billed based on the region the traffic leaves from and the continent of the connection location. That billing model favors putting attachments and their Cloud Routers in the regions where the workloads run. When each workload region has its own attachment, traffic goes directly between that region and the connection, with no inter-region data transfer charges. An attachment in a region with no workloads forces every packet to cross regions inside the VPC network first. That adds a hop, and the inter-region transfer is billed on top of the Interconnect egress.

Placing the attachment in the workload's own region also keeps routing simple. The on-premises routes its Cloud Router learns are already usable in that region, so regional dynamic routing is enough. Two constraints limit where attachments can go.

  • A Cloud Router can only manage VLAN attachments in its own region, so a router and its attachments always sit in the same region.
  • The attachments must be in a region that the interconnect location can serve.

Raising capacity on a saturated attachment

An attachment has its own provisioned capacity, separate from the capacity of the physical connection underneath it. When an attachment is saturated but the connection still has room, two approaches raise capacity without new circuits.

  • Edit the attachment's capacity in place. Dedicated Interconnect lets you change an existing attachment's capacity, so the attachment can use more of the room on the connection without new hardware.
  • Add a second attachment in the same region on the same connection. It gets its own BGP session with Cloud Router. When on-premises advertises the same prefixes with equal MED over both sessions, Cloud Router installs equal-cost routes and spreads many flows across both attachments with ECMP.

MED decides whether an additional attachment actually carries traffic. With equal MED values the attachments share the load, while an attachment advertised with a higher MED is a less preferred path that waits as a backup, which suits failover but adds no active capacity.

Taken together, these rules give a consistent order of reasoning for Professional Cloud Network Engineer scenarios. Put attachments where the workloads run, keep each Cloud Router in the same region as its attachments, and resize or add attachments on an existing connection before ordering new circuits.

Our Professional Cloud Network Engineer practice exams cover VLAN attachment placement and capacity alongside Cloud Interconnect SLA topologies and dynamic routing modes, with detailed explanations that drill these distinctions.

Get tips and updates from GCP Study Hub

arrow