Under-fetching
Under-fetching happens when insufficient data is requested, and thus additional round-trips are needed to satisfy all required data. This is often visualized as a request waterfall.
Symptoms
- Request amplification: repeated follow-up queries to fill in missing data.
- Increased latency as the network round trips in the waterfall take significant time (for client-server), or the serialization/deserialization costs add up (for server-storage).
- Progressive and slow rendering: data requirements for the next layer are only discovered after the previous layer has been fetched.
- Increased backend load from handling significantly more requests.
Why it matters
Under-fetching makes the backend do more work than necessary, increasing hosting costs. Request waterfalls can severely increase latency, giving users a bad experience.