toolbox
Home / AI, Data & Technology / Network Redundancy Bandwidth Requirements Estimator

Network Redundancy Bandwidth Requirements Estimator

The Network Redundancy Bandwidth Requirements Estimator calculates the per-link capacity needed to carry a protected traffic load after one or more parallel links become unavailable. Rather than sizing only for normal operation, it redistributes the full protected load across the surviving links and then adds a configurable headroom margin. This is useful for dual-carrier designs, link aggregation, redundant WAN paths, data-center uplinks, or any topology where traffic must continue during a defined link-failure scenario.

The estimate exposes a common redundancy tradeoff: adding more active paths can reduce normal per-link loading, but each surviving path still needs enough capacity to absorb redistributed traffic when failures occur. Use the result to compare proposed circuit sizes or to test whether an existing redundant bundle has adequate failover headroom. The model assumes the protected traffic can actually be rerouted across all surviving links; routing policy and unequal path capacities require additional design checks.

Calculator inputs

Mbps
links
links
%
Result
—
required capacity per surviving link
Surviving links—
Normal load per link—
Failover load per link—

1. Enter the protected traffic load
Use the aggregate Mbps that must remain serviceable during the modeled failure.

2. Specify parallel links
Enter the number of active links that share the protected load in normal operation.

3. Choose the failure tolerance
Set how many of those links may be unavailable at the same time. At least one link must remain.

4. Add operating headroom
Use a margin for bursts, growth, measurement uncertainty, or traffic not included in the protected-load figure.

5. Compare required link size
The main result is the minimum equal capacity each surviving link should have under the selected failure scenario and headroom assumption.

Surviving links = Active links − Failed linksFailover load per link = Protected traffic ÷ Surviving linksRequired link capacity = Failover load per link × (1 + Headroom ÷ 100)

Where:

  • Protected traffic — aggregate Mbps that must be carried after failure
  • Active links — equal parallel paths available in normal operation
  • Failed links — simultaneous unavailable paths in the design scenario
  • Headroom — extra capacity margin added after failover redistribution

Assumptions: Traffic is evenly distributable across equal-capacity surviving links and the selected failure count is the design case being protected against.

What the result means

The main result is the equal link capacity needed so the surviving paths can carry all protected traffic plus the chosen headroom.

Routing constraints, asymmetric circuits, oversubscription elsewhere, and convergence behavior can require additional capacity beyond this simplified model.

Given:

  • 2,400 Mbps protected aggregate traffic
  • 3 parallel active links
  • 1 simultaneous link failure
  • 20% headroom

Calculation:
Surviving links = 3 − 1 = 2. Failover load per link = 2,400 ÷ 2 = 1,200 Mbps. Required capacity = 1,200 × 1.20 = 1,440 Mbps.

Result:
1,440 Mbps required per surviving link.

Interpretation:
Each of the two remaining links needs at least 1.44 Gbps under this design case to carry the redistributed protected load with 20% headroom.

Why size for surviving links instead of normal load per link?

Redundant designs are expected to maintain service after a failure. Sizing only for normal sharing can leave the remaining paths overloaded exactly when redundancy is needed.

Can I model N+1 and N+2 designs?

Yes. Set the number of active links and the number of simultaneous link losses you want the design to survive. The failure count must be lower than the active-link count.

What if the links have different capacities?

This equal-link model will not represent unequal paths accurately. Model the constrained failure case separately or perform a path-by-path traffic allocation using the actual circuit capacities.

Should backup-only links be counted as active links?

Count them only if they are expected to carry the protected traffic in the modeled failover state. If a backup path has different routing or capacity, a separate scenario is safer.

Does headroom include future growth?

It can, but only if you intentionally choose the percentage for that purpose. You may prefer to forecast traffic growth first and then apply a smaller operational headroom margin to the forecast load.