Calculate aggregate bandwidth requirements for a network core given distribution layer demand.
Core network capacity planning must account for redundancy, since uplinks are typically deployed in redundant pairs or groups so the network survives a link or device failure — but that means some of the raw aggregate capacity can never be used for production traffic under normal conditions, because it must remain available to absorb the load if a sibling link fails. A 50% redundancy overhead models an N+1 or active/active design sized so that losing half the links still carries full traffic; lower percentages model designs with more links than the minimum needed for failover.
Usable capacity
capacity = uplinks × speed × (1 − redundancy_overhead%)
In an active/active design sized for N+1 resilience, every link carries live production traffic, so if any one fails its traffic must shift onto the survivors — sizing usable capacity at half of a two-link pool guarantees the remaining link can absorb 100% of the load without oversubscription during a failure.
Active/passive (one link idle as a hot standby) effectively has ~50% overhead built into the architecture regardless of link count, while active/active designs spread load across all links but must reserve capacity proportional to how many links can fail simultaneously and still meet demand — more links in the group generally lowers the required overhead percentage for the same failure tolerance.
Most enterprise core designs plan for a single link or device failure (N+1) as the baseline, reserving capacity accordingly — planning for simultaneous multi-link failure (N+2 or better) is typically reserved for mission-critical facilities where the cost of extra bandwidth headroom is justified by the availability requirement.