Delta Provider Read Scheduling

Delta Arrow Reader owns file-read concurrency, prefetching, Parquet reads, dynamic partition pruning, backpressure, cancellation, schema transforms, and deletion-vector masking.

For the execution flow, see Read scheduling. For exact reader defaults and metric definitions, see Execution options and Scan metrics.

Configure the reader through Delta Funnel

Python users pass a typed ProviderScanOptions object to Session. Rust users set the corresponding options through SessionOptions. Delta Funnel validates these values before it registers a table with the standalone reader.

Delta Funnel changes one standalone default: it uses ordinary Arrow Utf8 and Binary arrays because they perform better in its transformation-heavy workflows. Set use_view_types=True for a measured scan-heavy workload that benefits from Arrow view arrays.

See the API reference for the options and their Delta Funnel defaults.

Observe reads through Delta Funnel

Delta Funnel carries the reader's metrics into source reports, execution profiles, and one terminal Parquet I/O tracing event. These surfaces preserve the meaning of the underlying scan counters.

See the diagnostics reference for what appears in reports and tracing events. Use the standalone reader's metrics reference for the meaning and measurement boundary of each counter.