Latency, Bandwidth, and Throughput - And Why Everyone Confuses Them

After understanding what a Packet is, what a Connection is, and how protocols behave, we can finally ask the question that bothers almost every system: why is this slow?

The answer starts with distinguishing between three concepts that sound similar - but describe completely different problems.

Latency: How Long It Takes to Start Getting a Response

Latency is the waiting time. The moment between “I sent” and “something started arriving.”

This is a measure of responsiveness, not rate.

Latency is affected by:

• Logical distance on the network
• The number of layers and intermediaries
• Coordination between parties
• And queues that form along the way

Even if you only send one small Packet - Latency still exists.

Bandwidth: How Much Can Be Transferred in Parallel

Bandwidth describes capacity. How much information can be transferred in a given amount of time.

This is a potential, not a promise.

High bandwidth means: if everything is flowing - you can stream a lot.

But it doesn’t say:

• When the information will start arriving
• Or whether it will arrive on time

Throughput: What Actually Got Through

Throughput is the actual result. How much information successfully passed over time.

It’s affected by:

• Bandwidth
• Latency
• Packet loss
• Queues
• And the protocol chosen

Throughput is what you feel - not what’s guaranteed.

Why It’s So Easy to Get Confused

A system can have:

• High Bandwidth
• Reasonable Throughput
• And still feel slow

Why? Because high Latency makes every action feel “stuck,” even if a lot of information is moving overall.

That’s why many systems look powerful on paper - and disappoint in reality.

An Analogy

Think of a highway.

Latency is the time it takes to reach the entrance to the highway. Bandwidth is the number of lanes. Throughput is the number of cars that actually got through.

A highway with ten lanes doesn’t help, if the traffic jam is at the exit from the city.

The Systemic Implication

Optimizing without understanding the difference between these three concepts almost always addresses the wrong problem.

Sometimes you need to:

shorten Latency and not increase Bandwidth.

Sometimes you need to:

improve Throughput and not touch response time.

Without this distinction - every improvement is a guess.

Looking Ahead

Once you understand that Latency doesn’t just disappear on its own, the quiet factor that amplifies it emerges: queues.

In the next post we’ll dig into Queues - and why they form even when it seems like there are enough resources.