Enter Network Metrics
Submit the form to estimate throughput ceilings, compare competing constraints, and identify the most likely bottleneck source.
Example Data Table
| Scenario | Link Capacity | Measured Throughput | Offered Load | RTT | Loss | Queue Delay | Estimated Ceiling | Likely Bottleneck |
|---|---|---|---|---|---|---|---|---|
| Branch WAN peak hour | 1000 Mbps | 620 Mbps | 780 Mbps | 28 ms | 0.8% | 18 ms | 736.16 Mbps | Link Capacity Saturation |
| Long-haul backup traffic | 500 Mbps | 190 Mbps | 240 Mbps | 95 ms | 0.2% | 7 ms | 205.40 Mbps | TCP Window / RTT Limitation |
| Wireless office uplink | 300 Mbps | 122 Mbps | 160 Mbps | 24 ms | 2.2% | 16 ms | 129.80 Mbps | Loss / Retransmission Impact |
Formula Used
1. Effective CapacityEffective Capacity = Link Capacity × (1 - Protocol Overhead ÷ 100)
2. User DemandUser Demand = Active Users × Per User Demand
3. Peak DemandPeak Demand = max(Offered Load, User Demand) × Burst Factor
4. TCP Window-Limited ThroughputWindow Throughput = TCP Window Bits ÷ RTT Seconds
5. Observed Capacity EstimateObserved Capacity = Effective Capacity × (1 - Queue Penalty - Loss Penalty - Device Penalty - Error Penalty)
6. Estimated CeilingEstimated Ceiling = min(Effective Capacity, Window Throughput, Observed Capacity)
7. Capacity ScoreCapacity Score = max(Peak Demand ÷ Effective Capacity, Throughput ÷ Effective Capacity, Interface Utilization) × 100
8. Queue ScoreQueue Score = (Queue Delay ÷ RTT × 55) + (Jitter × 1.5)
9. Loss ScoreLoss Score = (Loss × 35) + (Retransmissions × 18) + (Physical Errors × 120) + Duplex Penalty
The highest score points to the dominant bottleneck domain. This model is diagnostic, not a packet capture replacement, so validate findings with live counters and traces.
How to Use This Calculator
- Enter the rated link speed and the measured throughput you observed.
- Add demand values, either from total offered load or user-based demand.
- Provide latency, queue, loss, retransmission, and interface utilization readings.
- Fill device health fields when routers, switches, or firewalls may be overloaded.
- Submit the form and review the primary bottleneck, severity, scores, and recommendations.
- Export the analysis as CSV or PDF for documentation or incident review.
FAQs
1. What does this calculator actually identify?
It estimates whether the main constraint comes from raw link capacity, TCP window and RTT, queue buildup, packet loss, retransmissions, or overloaded network hardware.
2. Can this replace packet captures and interface counters?
No. It is a triage tool. Use it to narrow investigation quickly, then confirm with switch counters, flow records, packet captures, and application telemetry.
3. Why does protocol overhead matter?
Headers, encapsulation, tunneling, and framing reduce usable payload capacity. Ignoring overhead can make a healthy link appear slower than expected.
4. When should I worry about TCP window limits?
Window limits matter when RTT is high and receive windows are too small. Long-distance transfers often suffer here, even when links have spare bandwidth.
5. What does a high queue score mean?
It means packets are waiting too long before transmission. This usually points to congestion, microbursts, poor scheduling, or oversubscribed uplinks.
6. How do retransmissions affect performance?
Retransmissions consume capacity without adding useful delivery. They also increase completion times and can amplify congestion during busy periods.
7. Why include device CPU and memory?
A firewall, router, or load balancer can become the choke point before the circuit does. High device load often causes drops, latency, and inconsistent throughput.
8. Is the highest score always the only problem?
Not always. Real bottlenecks can stack together. A saturated link can also create queues, and queues can trigger loss, so review the full score pattern.