
Dedicated Interconnect and Partner Interconnect both give private connectivity from an on-premises network to VPC networks over internal IP addresses, without crossing the public internet. The difference is who supplies the physical link. With Dedicated Interconnect, your own equipment connects directly to Google in a colocation facility. With Partner Interconnect, a supported service provider that already connects to Google carries the traffic. That difference shapes cost, lead time, BGP responsibilities and the provisioning workflow, and each of those comes up on the Professional Cloud Network Engineer exam.
Dedicated Interconnect is a direct physical connection between your equipment and Google. It requires your own routers in a Google colocation facility, a cross-connect to Google's equipment, and circuits of 10 Gbps or 100 Gbps. Your router runs BGP with Cloud Router. It gives the lowest-latency path, but provisioning usually takes weeks, and the connection carries port and cross-connect charges. One connection can serve many teams. A single Dedicated Interconnect connection with VLAN attachments in a shared host project can give many project teams private access to the data center.
Partner Interconnect uses a supported service provider that already maintains its own connections to Google in the metro. It fits when no colocation facility is nearby, when you have no equipment in one, or when the needed capacity is small. Providers offer two kinds of service, and the choice determines who runs BGP with Google.
Layer 3 service is the fit when an edge appliance supports only static routing and can't speak BGP, since with Dedicated Interconnect or Layer 2 service the customer router runs BGP itself. Partner Interconnect VLAN attachments depend on BGP through Cloud Router, so static routes on the attachments can't replace the dynamic sessions.
Both Interconnect types depend on BGP sessions with Cloud Router. BGP graceful restart is the feature that keeps traffic moving while a session restarts: a router that uses it keeps the routes it learned from its peer and keeps forwarding while the session comes back up. Cloud Router supports graceful restart, but the feature only helps when the peer on the other side supports it as well.
After Google approves and provisions a Dedicated Interconnect order, it creates a Letter of Authorization and Connecting Facility Assignment (LOA-CFA) for each connection. Google emails these documents to the NOC contact address supplied when the order was placed, and they can be downloaded at any time from the Interconnect page under Network Connectivity in the Google Cloud console. You then pass the LOA-CFA to the colocation facility operator, who uses it to run the cross-connect to the port Google assigned. Dedicated Interconnect uses the LOA-CFA for this step, not a pairing key.
Getting the LOA-CFA to the facility operator is the customer's job, which is why the documents go to the customer's own NOC contact.
Partner Interconnect provisioning begins in your own Google Cloud project. There is no physical link to order, because a supported provider already has connectivity to Google in the metro.
Dedicated Interconnect fits when the company has routers in a Google colocation facility, needs 10 Gbps or 100 Gbps circuits or the lowest latency, and can wait weeks for provisioning. Partner Interconnect fits when a colocation facility is out of reach, when the company has no equipment there, or when the capacity needed is small. On the Professional Cloud Network Engineer exam, the routing capability of the on-premises device is a second filter, because it decides whether the customer router or the provider runs BGP with Cloud Router.
Our Professional Cloud Network Engineer practice exams cover Dedicated and Partner Interconnect provisioning alongside VLAN attachment placement and Cloud Router ASN rules, with detailed explanations that drill these distinctions.