Throughput, Bandwidth, and Latency
Bandwidth, throughput and latency are three distinct measures of network performance. Bandwidth is the theoretical maximum data a connection can carry per second, throughput is the rate actually achieved after overhead, congestion and retransmission, and latency is the time a packet takes to travel from source to destination, usually in milliseconds.
Bandwidth, throughput, and latency are three related but genuinely distinct measurements of network performance, and confusing them leads to misdiagnosing real network problems, since the fix for each is completely different.
Bandwidth
The theoretical maximum amount of data a connection can carry in a given period of time, usually measured in bits per second (Mbps, Gbps). It's a ceiling set by the physical medium and equipment, not a guarantee of actual achieved speed.
Throughput
The actual, real-world rate of successful data transfer over a connection. Throughput is almost always lower than bandwidth, because of protocol overhead (headers, acknowledgments), network congestion, packet loss and retransmission, and other real-world inefficiencies that eat into the theoretical ceiling.
Latency
The time it takes for a single unit of data (a packet) to travel from source to destination, usually measured in milliseconds. Latency is largely determined by physical distance and the number of intermediate hops, not by how much data capacity is available.
Why these three don't move together
A connection can have high bandwidth but high latency: a satellite internet link can carry a lot of data per second (high bandwidth), but each individual signal still has to travel an enormous physical distance to the satellite and back, so the delay before the first byte arrives is genuinely large (high latency), no amount of bandwidth fixes that physical distance.
A connection can also have high bandwidth but low throughput: a home internet plan advertised at 500 Mbps might only achieve 350 Mbps in practice due to network congestion during peak hours, overhead from the specific protocols in use, or WiFi interference, none of which change the plan's rated bandwidth, only what's actually achieved.
A worked example
Two scenarios with a video call:
High bandwidth, high latency: A satellite internet connection with 100 Mbps bandwidth but 600ms latency. Video quality can be good (plenty of bandwidth for the video data), but there's a noticeable, frustrating delay between someone speaking and the other person hearing it, caused by latency, not by insufficient bandwidth.
Lower bandwidth, low latency: A fiber connection with only 20 Mbps bandwidth but 10ms latency. Video quality might need to be capped lower (less bandwidth available), but the conversation feels natural and responsive, with no noticeable delay, because latency is low.
Where this matters in real systems
Diagnosing "my internet feels slow" requires figuring out which of these three is actually the problem, and they call for entirely different fixes: low bandwidth means upgrading the plan or connection type; low throughput despite adequate bandwidth points toward congestion, interference, or an overloaded router; high latency, especially for something physically far away or satellite-based, often can't be fixed by upgrading bandwidth at all, since it's fundamentally a distance and hop-count problem. This is exactly why gamers and video-call users often care more about latency ("ping") than raw bandwidth numbers, since a fast-paced interactive activity is far more sensitive to delay than to raw data volume.
What interviewers actually ask about bandwidth, throughput, and latency
- "What's the difference between bandwidth and throughput?" Bandwidth is the theoretical maximum capacity of a connection; throughput is the actual achieved rate of data transfer, usually lower due to overhead, congestion, and retransmissions.
- "Can a connection have high bandwidth and high latency at the same time?" Yes, a satellite link is the classic example: it can carry a lot of data (high bandwidth) but each signal still takes a long physical round trip to the satellite and back (high latency).
- "Why might real-world throughput be lower than a connection's rated bandwidth?" Protocol overhead, network congestion, packet loss requiring retransmission, and interference (for wireless connections) all reduce actual achieved throughput below the theoretical ceiling.
- "Why do gamers care more about latency than bandwidth?" Fast-paced, interactive activity is highly sensitive to delay (a laggy response feels bad even with plenty of data capacity available), while raw data volume needs for something like a game's input data are relatively small.
Key takeaways
- Bandwidth is the theoretical maximum data capacity of a connection; throughput is the actual achieved rate, usually lower.
- Latency is the delay for a single packet to travel from source to destination, driven mainly by physical distance and hop count.
- These three measurements can vary independently: high bandwidth doesn't guarantee low latency, and rated bandwidth doesn't guarantee matching real-world throughput.
- Diagnosing network performance problems requires identifying which of the three is actually the bottleneck, since the fix for each is different.
More in Physical Layer
Browse all Computer Networks topics
Log in to save which topics you've read across every subject.
Frequently asked questions
What's the difference between bandwidth and throughput?
Bandwidth is the theoretical maximum capacity of a connection; throughput is the actual achieved rate of data transfer, usually lower due to overhead, congestion, and retransmissions.
Can a connection have high bandwidth and high latency at the same time?
Yes, a satellite link is the classic example: it can carry a lot of data (high bandwidth) but each signal still takes a long physical round trip to the satellite and back (high latency).
Why might real-world throughput be lower than a connection's rated bandwidth?
Protocol overhead, network congestion, packet loss requiring retransmission, and interference (for wireless connections) all reduce actual achieved throughput below the theoretical ceiling.
Why do gamers care more about latency than bandwidth?
Fast-paced, interactive activity is highly sensitive to delay (a laggy response feels bad even with plenty of data capacity available), while raw data volume needs for something like a game's input data are relatively small.