NETWORK RESILIENCE
Capacity is a failure scenario.
A healthy utilization chart describes today. A useful capacity plan also describes what happens when something disappears.
Ask where the traffic goes next
Capacity planning becomes more useful when it is attached to a concrete event. A link fails. A region enters maintenance. A service withdraws an announcement. The question is then specific: where does the affected traffic move, and can that destination carry it?
A small example
Imagine two independent links, each with 100 units of usable capacity, carrying 60 units apiece. In normal operation, each is at 60 percent. If all traffic moves onto one surviving link, demand becomes 120 units against 100 available. The original utilization looked comfortable, but this simplified failure scenario has a 20-unit shortfall.
Look beyond the total
Real networks have more constraints than that example. Traffic may not redistribute evenly. A downstream interface, a shared dependency or the service itself may become the limiting factor. Capacity that exists elsewhere is useful only if traffic can reach it and the receiving system can handle the work. State those assumptions before relying on the number.
Plan the operating response
For each important scenario, write down the expected traffic movement, the limiting resource and the action an operator can take. Include how to recognize that the prediction was wrong. Decide what can be restored quickly and what requires procurement or architectural change. That turns headroom from a reassuring percentage into an explicit statement about which events the service is prepared to handle.