Why Iceberg
Apache Iceberg is an open table format for large analytical datasets. Instead of a proprietary export, feed data is organized as tables that Iceberg-aware engines can read in place:- No ETL to maintain. You query the data where it lands rather than copying it into your warehouse first.
- Engine choice. Iceberg is supported across the modern data stack, so the same tables serve SQL warehouses, Spark jobs, and local notebooks.
- Fresh data. Data is committed as it lands, so analytical reads see recent blocks without waiting on a batch export.
Compatible platforms
Iceberg-compatible workflows cover platforms and tools including:Early access. Integration availability for specific platforms can vary, confirm the platforms and
workflows you plan to use when you onboard. See Early access.
Typical workflows
- Warehouse SQL: point Snowflake, BigQuery, or Databricks at feed tables and join onchain activity against your own data.
- Notebook analysis: pull feed data into DuckDB or Pandas for exploration and backtesting.
- Batch pipelines: run Spark or Trino jobs over full history for aggregation and feature engineering.
- Modeled marts: build dbt models on top of feed tables and ship curated datasets to your team.