- Bandwidth: maximum capacity of a link (bits per second, for example 100 Mbps).
- Throughput: the actual rate achieved, always at or below bandwidth.
- Latency (delay): time for data to travel from source to destination. Made up of propagation, transmission, queuing and processing delay.
- Jitter: variation in latency, harmful to voice and video.
- Error rate: proportion of bits received incorrectly.
Worked example: a 10 MB file is sent over an 8 Mbps link. 10 MB = 80 Mb, so transmission time = 80 / 8 = 10 seconds (ignoring propagation delay and overhead).
Bytes or bits? File sizes are measured in bytes (B) but link speeds in bits per second (b). One byte is 8 bits, so multiply a file size by 8 before dividing by the link speed.
Bandwidth and throughput: the motorway analogy
Think of a link as a motorway. Bandwidth is the number of lanes: the most traffic it could ever carry. Throughput is how many cars actually get through per hour. Throughput is always lower than bandwidth, because of protocol overhead (headers that are not your data), sharing the link with other users, errors that need resending, and the speed of the slowest link on the path. A 100 Mbps Internet plan might give a download throughput of 85 to 95 Mbps on a good day.
The four parts of latency
| Delay | Cause | Depends on |
|---|---|---|
| Propagation | The signal travelling along the medium, at nearly the speed of light | Distance |
| Transmission | Putting all the bits of a packet onto the link | Packet size ÷ bandwidth |
| Queuing | Waiting in a router's buffer behind other packets | How busy the network is |
| Processing | The device reading the header and deciding where to send the packet | Device speed |
Latency is usually measured as a round-trip time (RTT): the time for a message to go and the reply to come back. The ping command shows it in milliseconds. Typical values: a few ms inside a LAN, 20 to 60 ms to a server in the same country, and 150 to 300 ms to the other side of the world.
Latency and bandwidth are different. A faster plan (more bandwidth) makes large downloads finish sooner, but it does not make a game feel more responsive. Responsiveness depends on latency, and most of that is the distance to the server.
Jitter and error rate
Jitter is the variation in delay. If voice packets arrive 20 ms, 90 ms, 25 ms and 70 ms apart, the sound breaks up even though the average is fine. Applications use a small buffer to even it out, at the cost of a little extra delay.
The bit error rate compares bits received wrongly with bits sent. Fibre has an extremely low error rate; Wi-Fi in a crowded area has a much higher one. Errors force retransmissions, which reduces throughput.
A second worked example
A 2 GB video is downloaded over a 50 Mbps link. 2 GB = 2,000 MB = 16,000 Mb (multiply by 8). Time = 16,000 ÷ 50 = 320 seconds, about 5.3 minutes. (Storage sizes are sometimes counted with 1 GB = 1,024 MB; for estimates, 1,000 is close enough.) If the measured throughput is only 40 Mbps, the time becomes 400 seconds. Always work with the same units: bits with bits per second.
Try your own numbers in the calculator below.
Putting the numbers together
When users say "the network is slow", measure before guessing:
- Throughput — how many bits per second you actually move (speed test, file copy timing).
- Latency — how long a round trip takes (
ping). - Jitter — how much latency varies (hurts calls more than average delay).
- Loss — percent of packets that never arrive (even 1–2% damages VoIP and games).
Bandwidth is the capacity of the pipe; throughput is what you got. Congestion, Wi-Fi interference, server load or a duplex mismatch can make throughput far below the link bandwidth.
Everyday example: a 100 Mbps link with 80% utilisation left for you in theory might deliver only 20 Mbps to one PC because Wi-Fi is shared, the AP is far away, or the server itself is slow.
Remember
- Bandwidth = capacity; throughput = measured rate; latency = delay; jitter = uneven delay.
- Real-time traffic needs low latency, low jitter and low loss — not only high bandwidth.