Northwind Electronics has two Virtual WAN hubs: one in West Europe and one in East US. Hosts in West Europe need to initiate TCP connections to services in East US spokes over the Azure backbone. Current configuration has hubs in the same Virtual WAN but traffic takes indirect routes. What must be true to allow hub-to-hub traffic to traverse the Virtual WAN global transit and reach spokes in the other region without creating VNet peering between regions?

For Virtual WAN hub-to-hub traffic to traverse the Azure backbone, both hubs must belong to the same Virtual WAN instance. The key is ensuring the destination hub knows how to reach the source hub's spoke prefixes. This is achieved by enabling route table propagation on the destination hub for the source hub's connected routes. This allows the Virtual WAN global transit fabric to correctly forward traffic between spokes in different regions without requiring manual VNet peering. Creating a Site-to-Site VPN between hubs is unnecessary and bypasses the Virtual WAN's global transit capabilities. Global VNet Peering between hub VNets is also not the native Virtual WAN solution for inter-hub connectivity. ExpressRoute Global Reach is for on-premises to on-premises connectivity via Azure, not for inter-hub spoke-to-spoke routing within Virtual WAN.