Fleet location ingestion
Design the real-time location ingestion backbone for a fleet platform tracking 5 million connected vehicles, each emitting a GPS+telemetry packet every 1 second (so ~5M writes/sec sustained). You must support (a) a live map showing any vehicle's current position with <2s staleness, (b) replaying a vehicle's exact route over the last 90 days, and (c) fleet-wide queries like 'all trucks currently inside this depot polygon'. Discuss the ingestion path, the storage split, partitioning to avoid hotspots, and the durability vs cost trade-off for 90 days of per-second history.
Clarify scale and constraints first. Propose a clean component breakdown, then go deep on the hard parts (data model, bottlenecks, consistency, failure modes) and name the trade-offs you are making.