Packet Loss Impact Calculator
Calculate how packet loss affects TCP throughput, file transfer time, voice quality and application requests, and see why a loss rate that looks negligible is not.
Last reviewed by the Radiatus Cloud team
Need this monitored 24/7?
Radiatus delivers managed cloud & network operations.
One percent loss is not one percent slower
TCP treats loss as congestion and reduces its window sharply, then recovers slowly, so the throughput cost of loss is far out of proportion to the number of packets lost. The Mathis formula bounds throughput at roughly the maximum segment size divided by the round trip time, multiplied by one over the square root of the loss rate. At one percent loss on a 100 millisecond path that ceiling is about 1.2 megabits per second regardless of whether the link is 10 megabits or 10 gigabits.
Loss and latency multiply
The same loss rate is far more damaging on a long path, because every retransmission costs a full round trip and because the window takes proportionally longer to recover. A tenth of a percent loss on a local network is invisible; the same rate on a transatlantic path caps throughput at a fraction of the link. This is why a fast international circuit can deliver less than a slow domestic one, and why measuring loss matters more than measuring bandwidth on any long path.
Different applications fail at different rates
Voice using G.711 tolerates about one percent loss before quality degrades noticeably, and compressed codecs tolerate less because each lost packet carries more speech. Video conferencing shows visible artefacts from about half a percent. A bulk file transfer merely slows down. An application making many small sequential requests suffers most of all, because each lost packet stalls that request for a retransmission timeout while the others wait.
Related tools
- Port Scanner — Check which common ports respond on a host and understand what each exposed service means. Scan only systems you are authorised to test.
- DNS Misconfiguration — Analyze DNS records (TXT, SPF) for security gaps.
- TLS Compat Tester — Check client-server compatibility for TLS versions.
- Firewall Rule Risk — Evaluate firewall rules for over-permissive exposure.
Frequently Asked Questions
Why does a small loss rate cost so much throughput?
Because TCP interprets any loss as congestion and halves its window, then grows it back slowly. The Mathis formula bounds throughput at the segment size over the round trip time divided by the square root of the loss rate, so the cost scales with the square root rather than linearly.
What loss rate is acceptable?
Under 0.1 percent for anything performance sensitive, under 1 percent for voice using G.711 and less for compressed codecs, and effectively zero for storage protocols. Above 2 percent most interactive applications are visibly affected.
Does loss matter more on a long path?
Considerably. Every retransmission costs a full round trip and the congestion window takes proportionally longer to recover, so the same loss rate is far more damaging at 200 ms than at 2 ms. This is why measuring loss matters more than measuring bandwidth on a long path.
Which congestion control handles loss best?
BBR, because it models bottleneck bandwidth and round trip time rather than treating loss as congestion. On a path with random loss rather than congestion loss it substantially outperforms CUBIC and Reno, both of which halve the window on any loss.
How do I distinguish congestion loss from random loss?
Congestion loss rises with offered load and appears alongside increasing latency as queues fill. Random loss stays roughly constant regardless of load and does not move latency. MTR over time shows which pattern you have.
Privacy & Security
Everything runs in your browser; nothing is uploaded.
How to Use
Enter loss rate, latency and bandwidth to see the impact on throughput and application behaviour.
Disclaimer: This tool is provided "as is" without warranty of any kind. Results are for educational and utility purposes.
Related Tools
Port Scanner
NetworkCheck which common ports respond on a host and understand what each exposed service means. Scan only systems you are authorised to test.
DNS Misconfiguration
NetworkAnalyze DNS records (TXT, SPF) for security gaps.
TLS Compat Tester
NetworkCheck client-server compatibility for TLS versions.