Calculator Inputs
Tip: leave measured throughput blank when you only want a modeled estimate. Add a target throughput when you want the page to suggest required raw bandwidth.
Example Data Table
| Scenario | Bandwidth | RTT | Loss | Streams | Payload / Overhead | Estimated Throughput | Notes |
|---|---|---|---|---|---|---|---|
| Office WAN | 100 Mbps | 35 ms | 0.2% | 4 | 1460 / 58 Bytes | 84.7 Mbps | Healthy efficiency with mild loss. |
| Long-haul Link | 500 Mbps | 120 ms | 0.8% | 6 | 1448 / 72 Bytes | 148.3 Mbps | Loss and delay reduce usable rate. |
| Data Center East-West | 10 Gbps | 2 ms | 0.01% | 2 | 9000 / 78 Bytes | 9.69 Gbps | Large frames improve efficiency. |
| Home Broadband Test | 300 Mbps | 18 ms | 0.5% | 3 | 1460 / 58 Bytes | 205.6 Mbps | Moderate overhead and consumer congestion. |
Formula Used
1) Protocol Payload Efficiency
Payload Efficiency = Payload Bytes ÷ (Payload Bytes + Overhead Bytes)
2) Nominal Payload Throughput
Nominal Throughput = Bandwidth × Utilization Cap × Payload Efficiency
3) Window-Limited Throughput
Window Ceiling per Stream = (TCP Window Size × 8) ÷ RTT
4) Loss-Limited Throughput
Loss Ceiling per Stream ≈ (1.22 × MSS × 8) ÷ (RTT × √Loss Probability)
5) Estimated Usable Throughput
Estimated Throughput = Minimum of nominal payload rate, total window ceiling, and total loss ceiling
How to Use This Calculator
- Enter the raw link bandwidth and choose its unit.
- Add payload and overhead values to model packet efficiency.
- Enter RTT, packet loss, TCP window, and stream count.
- Set a utilization cap to reflect policy or practical usage.
- Optionally enter measured throughput to compare reality with the model.
- Optionally enter a target throughput to estimate required bandwidth.
- Add a transfer size if you want a transfer-time estimate.
- Press the calculate button and review the result table and chart above the form.
Frequently Asked Questions
1) What is the difference between bandwidth and throughput?
Bandwidth is the maximum raw link capacity. Throughput is the useful data rate you actually deliver. Overhead, delay, loss, window size, and congestion usually make throughput lower than bandwidth.
2) Why can a 1 Gbps link deliver far less than 1 Gbps?
The line rate includes headers, framing, retransmissions, idle periods, and protocol behavior. High latency, small windows, packet loss, and server limitations can reduce usable application throughput dramatically.
3) Why does packet loss matter so much?
TCP reacts to loss by reducing its sending rate. Even small loss percentages can heavily limit throughput, especially on long-distance paths with large round-trip times.
4) What does the TCP window input represent?
The TCP window controls how much data can stay in flight before acknowledgments return. If the window is too small for the delay, the sender cannot fill the path efficiently.
5) Why are parallel streams included?
Multiple flows can sometimes overcome per-stream limits, especially when one stream is window-limited. They do not remove protocol overhead, and they still share the same raw bandwidth ceiling.
6) Is the loss formula exact?
No. It is a useful approximation based on classic TCP behavior. Modern congestion control, buffering, ECN, and application design can produce results above or below this estimate.
7) Should I use decimal or binary units?
Use whichever matches your environment. Network rates are commonly quoted in decimal units like Mbps. Memory and buffer sizes are often described with binary units like MiB.
8) Can this calculator help with capacity planning?
Yes. It helps compare advertised link speed with expected usable performance, estimate transfer times, and gauge how much bandwidth may be needed for a practical throughput target.