Transit must continuously append AVL vehicle positions into a bronze Delta path. Which technology should they build on?

Transit continuously appending AVL vehicle positions into a bronze Delta path should build on Spark Structured Streaming. Structured Streaming is the continuous ingest engine for civic telemetry that must keep appending positions as vehicles move, writing into Delta bronze for downstream ETA, bunching, and operations analytics. Continuous append into bronze is the defining requirement; batch snapshots cannot substitute for a live AVL feed. A scheduled Spark batch job to append AVL vehicle positions into bronze Delta is not continuous ingest; it appends in scheduled intervals with lag instead of appending each vehicle position as it arrives. Spark Structured Streaming to append AVL vehicle positions into bronze Delta view misses the required sink; a view is a logical definition, not the bronze Delta path, so it never persists raw positions into bronze storage. A Databricks SQL warehouse to append AVL vehicle positions into bronze Delta is the wrong engine for continuous ingest; it is a serving layer for analyst queries, not an ingest sink, so it cannot keep appending the live AVL feed to the bronze path.

Exam caveat: Structured Streaming into Delta bronze is the platform pattern—still configure checkpoints, triggers, and schema for the AVL feed; “streaming” without a sink and checkpoint plan is incomplete for production transit ops. Operational check: create a Structured Streaming query that reads AVL positions and appends to bronze Delta with checkpointing, run for a soak period, verify file growth and position freshness, confirm a restart resumes without rebuilding the entire history by hand, and alert on stream lag.