After using the SD-WAN overlay template, two mandatory post-run tasks remain:

"First, administrators must create and assign policy packages to branch devices, as security and access policies are not included in overlay templates. Second, SD-WAN rules must be configured so that traffic can be matched and steered appropriately through the established overlays. Neglecting either task results in ungoverned traffic or inefficient routing, undermining the benefits of SD-WAN."

Templates automate topology, but policy and rule definition are critical for operational effectiveness.

When you enable ADVPN (auto-discovery VPN) in the overlay template, FortiManager automatically updates both the IPsec and BGP templates on the hub so that shortcut tunnels can be established dynamically.

ADVPN can be activated in the SD-WAN overlay template for any supported topology, including dual-hub primary–primary, not just single hub.

Enforce Device Configuration is enabled and the blueprint applies the provisioning CLI templates. The LAN-interface script sets port1 and port2 to DHCP and assigns a static IP to port5 (using the branch_id variable). Therefore, when FortiManager pushes the blueprint, it updates the configurations of port1, port2, and port5 - and their IP addresses may change accordingly.

The router policy explicitly denies traffic with source 10.0.1.128/25 (which includes 10.0.1.130) and destination 128.66.0.0/24 (which includes 128.66.0.125). Even though SD-WAN service 4 shows members (port1 and port2) alive and available for this traffic, the router policy is evaluated first and blocks it. Therefore, FortiGate drops the traffic flow.

For SD-WAN deployments that span multiple regions—where most traffic is intra-region and there is no requirement for inter-region ADVPN—the best practice is to use IBGP with BGP on loopback interfaces for routing within each region and EBGP between the regions. This approach ensures robust and scalable routing, isolates regional routing domains, and enables policy control at region boundaries. BGP on loopback is preferred for its reliability and flexibility, as it enables peering that is not tied to specific physical interfaces. EBGP between regions allows each region to maintain independent routing policies and summarization, optimizing performance and manageability. By separating IBGP (intra-region) and EBGP (inter-region), you create a modular architecture that scales easily and simplifies fault isolation and troubleshooting.