What are the components of latency, and roughly how big is each one?
Show the answer
Say this first
Four parts. Propagation: distance divided by signal speed; light in fiber covers about 200 km per millisecond, so 5 microseconds per km. Serialization: frame size divided by link rate; 1500 bytes takes 12 microseconds at 1 Gbps but about 8 ms on a 1.5 Mbps T1. Queuing: time waiting behind other packets, the variable part that creates jitter. Processing: the forwarding lookup, microseconds in hardware.
Go deeper
Worked numbers
| Component | Example | Value |
|---|---|---|
| Propagation | 3,000 km of fiber path | about 15 ms one way, 30 ms RTT minimum |
| Serialization | 1500 bytes at 1 Gbps | 12,000 bits / 10^9 = 12 microseconds |
| Serialization | 1500 bytes at 10 Gbps | 1.2 microseconds |
| Serialization | 1500 bytes at 1.544 Mbps | about 7.8 ms |
| Queuing | A 1 MB buffer draining at 1 Gbps | up to 8 ms |
| Processing | Hardware forwarding | a few microseconds |
What you can and cannot change
- Propagation only changes with a shorter path (better routing, a closer region).
- Serialization matters on slow links. It is why voice over low-speed WAN needed fragmentation and interleaving.
- Queuing is where QoS, buffer tuning and avoiding congestion help. Deep buffers that stay full cause bufferbloat: high latency without loss.
They will ask next
Why is the RTT from Europe to the US east coast never below about 70 ms?
Fiber paths across the Atlantic are around 6,000 to 7,000 km one way once you include the cable route and terrestrial legs. At 5 microseconds per km that is about 30 to 35 ms each direction before any queuing. Physics sets the floor.
Traps
- Saying a faster link reduces propagation delay. It only reduces serialization.
How did you do?Book 1.5 Performance