Class DerivedTableWriter

java.lang.Object
io.deephaven.enterprise.derivedtablewriter.DerivedTableWriter

public class DerivedTableWriter extends Object
Persist data from a Table to Deephaven storage using an in-worker DataImportServer.

This utility is intended to read one live table from storage, perform query operations, and then write a derived live table to storage.

The source table must be either a blink table or an add-only table. Each new row is persisted by the DIS.

To ensure exactly-once delivery, your input table must have a column of ascending sequence numbers specified using the DerivedTableWriter.Options.sequenceColumn. If you do not have a sequence column, then upon restart rows may be logged twice. If your source table is ephemeral, then even with a sequence number rows may be dropped on restart. As an alternative, you may also read the derived table and extract a sequence number, timestamp, or other information that can be used to resume your source appropriately before starting the DerivedTableWriter.

When processing independent partitions, using the ingestPartitionedTable(String, String, DataImportServer, PartitionedTable, PartitionedOptions, Options) is recommended. Each constituent is independently logged as an output partition.