A public hazard desk mirrors a global seismic feed for a response map. Thousands of browsers keep the world view open. Operators also need to inspect the detailed products for one selected event and reconstruct what the desk saw before a later source revision.
The supplied fixed USGS teaching snapshot contains 86 point events and its delivered GeoJSON file is 21,932 bytes. Use that measured payload only as a sizing proxy; the live USGS feed's count and size vary. USGS publishes summary feeds on a one-minute refresh cadence and exposes details for selected events. For the exercise, the desk plans for 20,000 concurrent viewers, each refreshing once per minute, for a 24-hour day. Compute direct_requests_daily and direct_transfer_gb for the naive case where every viewer fetches one uncompressed 21,932-byte summary on every refresh. Use decimal GB (10⁹ bytes); ignore headers and compression in that estimate.
Design the read paths for the global overview and a clicked event, the upstream ingestion, and how to retain provenance when a source event is revised. The overview includes all current events in this small feed. Choose one option per decision and justify it. Justifications are recorded for review but not machine-scored; the choices and two calculations are scored.
The 86-event file is a fixed M5.5+ January–March 2025 snapshot, not a claim that any live minute feed always contains 86 events. Model and monitor the real feed size before setting production budgets.
These are where the teaching is. Read them twice.
USGS GeoJSON summary feeds are updated every minute and carry event IDs and detail links (https://earthquake.usgs.gov/earthquakes/feed/v1.0/geojson.php); the detail document includes per-event products (https://earthquake.usgs.gov/earthquakes/feed/v1.0/geojson_detail.php). The 86-feature, 21,932-byte proxy is measured from the Kharita fixed teaching snapshot.
The decision is graded. What you write beside it is kept and shown next to the official reasoning when you are done — it is never scored.
How should the service obtain each new upstream summary?
What should serve the small, global point overview to viewers?
How should a selected event expose its detailed products?
How should a revised origin or magnitude be represented for audit?
Work the problem in whatever tool you like, then enter the answers here.
direct_requests_dailydirect_transfer_gb6 scored, 0 informational
Shared upstream ingestion
20%Overview delivery for a small global layer
20%Event details on demand
13%Revised event identity and audit
20%Direct browser fetches per day
13%Uncompressed direct transfer per day
13%