When users complain about "slow internet," they are usually conflating several different network concepts. To troubleshoot effectively, network engineers must understand the exact differences between Bandwidth, Throughput, and Goodput.
Definitions
| Term | Definition | Example |
|---|---|---|
| Bandwidth | The theoretical maximum capacity or configured speed of the physical link. | A 300 Mbps ISP fiber connection. |
| Throughput | The actual, measured rate of total data successfully transmitted over the link. | A speed test maxing out at 285 Mbps. |
| Goodput | The actual rate of useful application payload delivered (excluding protocol headers and retransmissions). | A file downloading at 260 Mbps. |
Bits vs Bytes (The Ultimate Confusion)
Internet speeds are sold in bits per second (Mbps with a lowercase 'b'). However, web browsers and file transfers display speeds in Bytes per second (MB/s with an uppercase 'B').
8 bits = 1 Byte.
If you purchase a 100 Mbps internet connection, your maximum theoretical download speed is roughly 12.5 MB/s. (100 รท 8 = 12.5). If a user sees a file downloading at "11 MB/s" and assumes they are only getting 11% of their 100 Mbps connection, they are misinterpreting the units.
Why is Throughput Always Lower Than Bandwidth?
You will never achieve 300 Mbps of goodput on a 300 Mbps link. Expected limitations include:
- Protocol Overhead: Every single packet requires Ethernet, IP, and TCP headers. If you are downloading a 10 GB file, a noticeable percentage of the bandwidth is consumed just sending the headers that carry the file chunks.
- VPN Encryption Overhead: Wrapping traffic in IPsec encrypts the data but adds massive headers, drastically lowering the available throughput for the actual file.
- Wi-Fi Conditions: A laptop might have a "Bandwidth" connection rate of 866 Mbps to the Wi-Fi access point, but due to RF interference and half-duplex contention, actual throughput might be only 200 Mbps.
- TCP Behavior: If there is high Latency or Packet Loss, TCP windowing algorithms deliberately slow down the transmission rate, causing throughput to plummet far below the available bandwidth.
Troubleshooting Poor Throughput
A web-based speed test alone does not diagnose the cause of slow throughput. Use this workflow to isolate the bottleneck:
- Confirm Negotiated Speed: Check the physical switch port. Is the PC accidentally negotiating at 100 Mbps instead of 1 Gbps due to a Duplex Mismatch or a damaged cable?
- Test Wired First: Always bypass Wi-Fi. Run the test hardwired directly to the router to eliminate RF variables.
- Check Current Utilization: Look at your Network Monitoring graphs. If you have a 300 Mbps link, and another user is currently downloading an OS update at 250 Mbps, your throughput will be awful. The link is simply congested.
- Check CPU Bottlenecks: Can your firewall actually inspect traffic at 1 Gbps? Older firewalls with Deep Packet Inspection turned on might max out their CPU at 400 Mbps, artificially capping your throughput.
- Test LAN Independently: Copy a file between two servers on the same local switch. If it runs at gigabit speeds, the LAN is healthy, and the bottleneck is the Firewall, the WAN, or the ISP.
- Compare Multiple Servers: If downloading from Server A is slow, but downloading from Server B is fast, your network is fine. The bottleneck is the remote server.