A link that transmits but cannot receive never hears the BPDU telling it to block, and spanning tree loops. UDLD detects unidirectional links and shuts the port first. Covers normal and aggressive mode with real CML output.
An MST core touching a Rapid-PVST access layer is one of the most misunderstood seams in Layer 2. Build the boundary in a CML lab, see the CIST at work, and deliberately trigger the PVST simulation inconsistency so you can recognize it.
Migrate a live network from EIGRP to OSPF without dropping a packet: run both protocols side by side, then flip administrative distance one command at a time. The ships in the night technique with real lab output.
Two-point redistribution is where networks go to die. This article builds the classic CCIE failure for real in a CML lab, shows a route being re-injected into the protocol it came from, and then fixes it with the exact output at each step.
Tunnels up, neighbors up, routes missing: split horizon breaks EIGRP over DMVPN's mGRE cloud. Covers split horizon and next-hop-self, dual-hub design, and the dual-homed loop, with real CML output.
Offset lists add metric to specific routes on a specific interface in a specific direction, without touching delay for everyone else. See them steer spokes to HUB1 in a dual-hub DMVPN on IOS XE, with exact before and after metrics.
EIGRP summarization shrinks routing tables but hides detail. A leak map lets you advertise the summary and still punch through specific prefixes that must stay visible. Shown working on Cisco IOS XE with real DMVPN lab output.
EIGRP Site of Origin stamps routes with an extended community so dual-homed WAN sites do not learn their own prefixes back. Covers what SoO does, how it is configured, and what we actually found trying it on IOS XE.
Five OSPF faults built and broken for real in a CML lab, ticket style: the symptom, the two commands to compare, the actual output, and the fix for each one. OSPF rarely tells you what is wrong; this method finds it anyway.
OSPF compares route types before metrics, so a worse metric routinely wins. Prove the full O, O IA, E1, N1, E2, N2 ordering by injecting the same prefix six different ways into a live CML topology and watching each one take over.
area range, filter-list, distribute-list, summary-address: four OSPF filters that are not interchangeable. Side-by-side lab output shows which ones change the LSDB and which only touch the local RIB.
The OSPF forwarding address can point a Type 5 route at a next hop that never advertised it, or silently drop it from the RIB while the LSA sits in the database. This article covers the five conditions that set it and the failure it causes.