Accumulated query cost and runtime

FOR CONTEXTBetter data preparation means less query engine workFresh-data pathNewlywrittenFreshdatacontinuouslyQuery-ready dataOrderedraw dataOrdered pre-aggregatedQuery engineQuerycomputeFastanswerscontinuouslyDrill-downsInteractive aggs
Drag to explore

Accepted active-ingestion pair: serverless_baseline_full_20260918T170453Z. Allocated ingestion and MV refresh DBUs plus estimated predictive optimization; normalized query runtime at public list rates. Storage, producer VM, network, idle/minimum warehouse charges, control SQL and post-ingestion work excluded. Databricks raw_rows is client durability progress; matching does not establish exact query-visible rows or identical MV freshness. Query curves are step sums. Complete-ingest preparation cost is allocated proportionally by row progress; it is not a metered time series. Zero is retained in readouts and omitted on logarithmic axes.