Application-aware routing

Meridian's voice calls between Manchester and Leeds should use MPLS, but last month the MPLS provider had a bad day: latency tripled and calls were unusable while the internet path was fine. You make the overlay notice and react by itself. An SLA class defines what voice needs, an app-route policy prefers MPLS while it meets that class, and a WAN emulator in the MPLS path lets you break the circuit on demand. ICMP stands in for voice, because you can see each packet's round-trip time.

What this course leaves out. There is no SD-WAN Manager (vManage) in any of these labs. A virtual edge can't join the Manager's inventory without a signed device list from Cisco, which these labs don't have. So you won't see the Manager's GUI, dashboards, templates, Policy Builder or zero-touch onboarding. You work at the CLI of the controllers and edges, with a private lab CA and device lists typed by hand. You read loss, latency and SLA state at the CLI, not in the Manager's tunnel health views. The hosts have fixed addresses set at boot, because the DHCP server on these vEdge images can lock up the edge at start-up.