Present the demo

Use the receipts from the current deployment. Historical Azure observations through 2026-08-08 describe a different topology and do not prove the new FRR relay or inside-VIP paths.

Start with the route

In Azure US, three CEs peer with both local FRR relays. Both FRRs peer with both Azure Route Server endpoints. The relays export only the CE-learned application VIP /32. The client’s effective route table should show two FRR next hops. Select the configured subscription explicitly:

cd terraform
az network nic show-effective-route-table \
  --subscription "$(terraform output -raw azure_subscription_id)" \
  --resource-group "$(terraform output -raw resource_group_name)" \
  --name "$(terraform output -raw client_nic_name)" \
  --output json

The external application domain resolves to the BGP VIP for the routed test. The private application domain uses the ILB frontend at inside-subnet .10. Send each domain as the HTTP Host value; a bare-IP request need not match an XC load balancer. The ILB’s .11 frontend independently checks Site Console TCP/65500 with a nonfloating rule.

Demonstration order

1. Show three ONLINE Azure CEs with registered, MAC-bound interfaces.
2. Show six CE–FRR and four FRR–Route Server sessions per region, the VIP at both FRRs, and two effective client next hops.
3. Send HTTP requests from each region’s client through its BGP VIP and ILB application VIP. Show the same owned origin response and healthy console frontend.
4. Stop one CE, demonstrate route withdrawal and continuous traffic, then start it and show recovery. Repeat with one FRR to demonstrate one surviving next hop. Use the lifecycle’s sanitized failure and recovery receipts.
5. Show AWS’s three sites, six Connect peers, twelve sessions, workload/VIP responses, and recovery receipts.
6. Show the KVM CE’s provider-owned SLO and physical-LAN SLI, FRR route, and independent client response through its custom inside VIP. Verify the host bridge and default route remain intact.
7. Finish with the refresh-enabled zero-change plan from the second build.

The Canadian regional deployment is maintained independently in Canada Topology.