What explains cross-region latency in distributed systems?

What explains cross-region latency in distributed systems?

• A. Latency is eliminated by moving data to the cloud.
• B. The speed of light through fiber imposes a minimum latency between NYC and London.
• C. Latency is caused by encryption overhead.
• D. Latency is irrelevant if you use caching.

Answer: (B)

Propagation delay from the finite speed of light in fiber sets a hard lower bound on cross-region latency. Signals can’t travel faster than light, and in fiber the speed is about two-thirds of vacuum due to the glass’s refractive index. The distance between NYC and London is roughly 5,500 km, so even in an ideal path the one-way travel time is around 25–30 ms, with roughly 50–60 ms for a round trip. Real networks add routing hops, signaling, and processing delays, so actual latency is usually higher. This physics-based limit explains why cross-region access cannot be eliminated just by moving data to the cloud. Encryption overhead and caching affect latency in practice but don’t negate this fundamental propagation delay: encryption adds some CPU/crypto time, and caching can reduce repeated access time, yet data still must physically traverse the distance on first access or when cache misses occur. Moving data to the cloud doesn’t remove the distance-based delay either.

Propagation delay from the finite speed of light in fiber sets a hard lower bound on cross-region latency. Signals can’t travel faster than light, and in fiber the speed is about two-thirds of vacuum due to the glass’s refractive index. The distance between NYC and London is roughly 5,500 km, so even in an ideal path the one-way travel time is around 25–30 ms, with roughly 50–60 ms for a round trip. Real networks add routing hops, signaling, and processing delays, so actual latency is usually higher. This physics-based limit explains why cross-region access cannot be eliminated just by moving data to the cloud.

Encryption overhead and caching affect latency in practice but don’t negate this fundamental propagation delay: encryption adds some CPU/crypto time, and caching can reduce repeated access time, yet data still must physically traverse the distance on first access or when cache misses occur. Moving data to the cloud doesn’t remove the distance-based delay either.