Datalake
A Datalake stores data as files in object storage so many tools can read it, instead of copying it into one vendor’s warehouse first.
A data warehouse is optimized for SQL analytics on modeled tables. Skipprd's Datalake enables that: ELT pipelines land data in the lake; Apache Iceberg tables on those files are what warehouse engines and Skipprd SQL both query. Iceberg is the open table format on the Datalake — snapshots, schema, and SQL over Parquet — without locking the data in a closed engine.
Skipprd is not a warehouse product. Warehouses you already use (Athena, Snowflake, and others) can read the same Iceberg tables.
How Skipprd uses it
- ELT ingest writes a durable WAL, then compact to Iceberg Parquet in object storage.
- Skipprd catalog (
catalog.type: skippr) holds Iceberg table pointers. Self-hosted clustered runs use a DynamoDB catalog table. - Clustered query unions the Iceberg snapshot with live WAL over Flight SQL. That path is documented in maintainer architecture.
See also How Skipprd Works and the Iceberg output.