Networking Goodput Explained Simply | Generated by AI
Question: What does “goodput” mean in networking?
Answer:
Goodput refers to the actual useful data successfully delivered to the application layer (or end user) per unit of time in a network connection.
It measures only the application payload data that is:
- Correctly received
- In the right order
- Without errors
- Without duplicates
- After removing all protocol overhead and useless/retransmitted data
Goodput vs. related terms
-
Throughput
→ Total number of bits successfully transmitted over the link per second (includes protocol headers, retransmissions, etc.) -
Goodput
→ Only the useful application data that the receiver actually uses (excludes headers, control packets, retransmitted data, duplicate packets, etc.) -
Bandwidth
→ Theoretical maximum capacity of the link
Typical relationship (in decreasing order)
Bandwidth ≥ Throughput ≥ Goodput
Practical examples
| Scenario | Throughput | Goodput | Difference explanation |
|---|---|---|---|
| Clean HTTP/2 download, no loss | ~980 Mbps | ~940 Mbps | Mainly HTTP/2 + TLS headers |
| TCP connection with 5% packet loss | 600 Mbps | 380 Mbps | Many retransmissions consume bandwidth |
| Video streaming with lots of FEC | 25 Mbps | 18 Mbps | Forward Error Correction packets don’t carry payload |
| VPN tunnel (high overhead) | 850 Mbps | 620 Mbps | IPsec/ESP + inner IP + TCP/UDP headers |
| File transfer over very lossy Wi-Fi | 40 Mbps | 8 Mbps | Extremely high retransmission rate |
Summary – most common way people remember it
Goodput = Throughput – useless traffic
(useless = protocol headers + retransmissions + control packets + duplicates + padding + FEC overhead that didn’t help)
In performance-sensitive applications (file transfer, video streaming, gaming, real-time protocols), engineers care much more about goodput than raw throughput.
References:
- None required for this conceptual explanation