Home › Learn › Data platform architecture › DataTwin Studio › Delta Lake internals
Scenes: One batch, end to endDelta Lake internalsSpark internalsProblem LabData qualityPerformance stories · more coming: streaming, Kafka, Unity Catalog, dimensional modelling, cost
DataTwin Studio · Delta Lake internals
Delta Lake internals, recorded
What really happens on disk when you create, insert, update and delete in a Delta table. Each step shows the command, the table's rows, every Parquet file on disk (live, or kept only for time travel), the commit files in _delta_log, the table history and the Spark jobs. Six scenarios, from 'a table is files plus a log' to 'why VACUUM breaks time travel'.
real Recorded on local[2] Spark 3.5.1 + Delta Lake 3.2.0 on 2026-10-28. Nothing here is drawn by hand: the files, commits, errors and timings are what Delta actually wrote and reported.
Read alongside
Data platform architecture · Medallion architecture · Open table formats · The small-files problem · Idempotency · Where data costs come from