MidNetworking

Explain the difference between latency, bandwidth, and throughput. Why might a high-bandwidth link still feel slow?

What they are really testing: Whether you can diagnose performance beyond "add more bandwidth." The latency-versus-bandwidth distinction is where a lot of slow-app misdiagnosis happens.

A real interview question

Explain the difference between latency, bandwidth, and throughput. Why might a high-bandwidth link still feel slow?

What most people say

drag me

Bandwidth is speed and latency is delay; if it is slow you need more bandwidth.

The reflex to add bandwidth is the trap the question sets. Many slow experiences are latency-bound or loss-bound, where more bandwidth changes nothing. It also blurs throughput into "speed".

The follow-ups they ask next

  • An API call is slow only for users far from the region. What is going on and what helps?

    Distance-driven latency: more round trips times a higher RTT. Reduce round trips (batch, cache), put a CDN or edge/PoP closer, or add a regional deployment, not more bandwidth.

  • Why does a single TCP stream sometimes not fill a fast, high-latency link?

    The bandwidth-delay product: throughput is capped by window size / RTT. You need a large enough TCP window (or parallel streams) to keep the long pipe full.

What the interviewer is listening for

  • Cleanly separates the three terms
  • Knows latency-bound workloads ignore bandwidth
  • Mentions RTT/round trips, loss, or window size

What sinks the answer

  • "Just add bandwidth"
  • Conflates throughput and bandwidth
  • No notion of round trips or distance

If you genuinely do not know

Say this instead of freezing. Reasoning out loud from what you do know beats silence every single time, and a good interviewer is listening for exactly that.

Latency is [the delay for one packet, often RTT]; bandwidth is [the pipe capacity]; throughput is [what you actually achieve, often less]. A fat pipe feels slow when [the workload is latency-bound, many sequential round trips], so the fix is [fewer round trips, caching, or a CDN closer to users], not more bandwidth.

Keep going with networking

All 336 cloud engineer questions

Knowing the answer is not the same as recalling it under pressure

Sign in to send the questions you fumble to spaced recall, so they come back right before you would forget them, and learn the concepts behind them with hands-on labs.

Start free