Questions: Routing Convergence, Flapping, and Damping

Question 1

A BGP route flaps 10 times in 20 minutes, accumulating a penalty of 3000. The suppress threshold is 2000. What does the router do with this route?

• A. It withdraws the route from the routing table permanently until manually reset
• B. It continues advertising the route but doubles the update interval to reduce churn
• C. It suppresses the route — stops advertising it to BGP neighbors — until the penalty decays below the reuse threshold
• D. It reroutes traffic over the secondary path and waits for the primary path to stabilize on its own

Answer: (C)

When a route's accumulated penalty exceeds the suppress threshold, BGP flap damping suppresses the route: the router stops advertising it to neighbors. The penalty continues to decay exponentially with the configured half-life. Only when the penalty drops below the reuse threshold is the route advertised again. Crucially, the route is NOT permanently withdrawn — it returns automatically once the penalty decays. Option A is wrong because suppression is temporary. Option B is not how damping works.

Question 2

An operator observes that a destination prefix has been unreachable for 45 minutes despite the underlying link being stable for the last 40 minutes. What is the most likely cause?

• A. BGP convergence is inherently slow and 45 minutes is within the normal convergence window
• B. The route was suppressed by flap damping due to earlier instability, and the penalty has not yet decayed below the reuse threshold
• C. The router's routing table is full and has dropped the prefix entry
• D. BGP hold timers have expired and the session is being re-established

Answer: (B)

This is the classic failure mode of over-aggressive flap damping. After the link stabilized, the penalty was still above the suppress threshold, so the route remained hidden from neighbors even though the underlying link was healthy. The half-life determines how long suppression lasts — with a 15-minute half-life and a high penalty, suppression can persist for 30–60 minutes after the link stabilizes. This illustrates why damping parameters must be carefully tuned: overly aggressive thresholds trade availability for stability.

Question 3

A route that flaps exactly once will never be suppressed by BGP flap damping, even with very aggressive penalty parameters.

• T. True
• F. False

Answer: True

Flap damping is designed to suppress repeatedly unstable routes while leaving occasionally-flapping routes unaffected. A single flap adds a fixed penalty increment (typically 1000) that begins decaying immediately. For suppression to occur, the penalty must exceed the suppress threshold (typically 2000). A single flap can only reach that threshold if the penalty increment itself exceeds the threshold, which standard configurations do not allow. The mechanism is specifically designed this way — planned maintenance events that cause a one-time route withdrawal should not trigger suppression.

Question 4

BGP flap damping prevents route flapping by stabilizing the underlying link before advertising the route to neighbors.

• T. True
• F. False

Answer: False

Flap damping does not prevent flaps — it hides them. The link continues to go up and down; the local router's routing table still reflects these changes. What damping does is suppress the route advertisements sent to BGP neighbors, so those neighbors do not receive the repeated withdraw/re-announce updates. The instability itself is not fixed — it is merely concealed from the rest of the network. This distinction matters: operators must still fix the underlying link issue; damping is only a tool to prevent the instability from propagating churn across the internet.

Question 5

Explain the difference between routing convergence and routing churn, and why route flapping makes routing churn a global internet problem rather than a local one.

Model answer: Routing convergence is the process by which all routers agree on a consistent topology after a change — it is a transition from one stable state to another. Routing churn is the ongoing processing load on routers from a stream of update messages, particularly when those updates do not lead to a new stable state. Route flapping generates churn because each flap causes a withdraw message followed by a re-announce message that propagates outward through the BGP mesh. Every router that receives these updates must recalculate its forwarding table. A single flapping link in one autonomous system can generate thousands of update messages across hundreds of global routers, consuming CPU on core infrastructure that has nothing to do with the flapping link. The problem is global because BGP's update propagation is internet-wide.

The distinction between convergence (finite, goal-directed) and churn (ongoing, resource-consuming) is key to understanding why flap damping exists. Convergence is acceptable and necessary; churn is pathological. Damping is justified because it trades a delay in convergence (routes take longer to re-stabilize) for elimination of churn. The tradeoff only makes sense when the instability is genuine — which is why damping should be conservative and targeted at aggregate routes rather than specific prefixes.