First-class Iceberg table support built into the core engine. Not a bolt-on connector, not a third-party wrapper. Read, query, and manage Iceberg tables with the same SQL engine and the same performance you expect from Delta Lake.
No vendor lock-in. Choose the table format that fits your workload.
Most data platforms force a choice: pick one table format and live with it. When they add Iceberg support, it arrives as a limited connector with gaps in spec coverage, missing delete support, or read-only access. Delta Forge takes a fundamentally different approach.
The same SQL engine that powers Delta Lake queries also powers Iceberg queries. There is no secondary code path, no reduced feature set, no "lite" mode. You get the full capabilities of both formats in a single platform.
Complete V1/V2/V3 implementation with no gaps and no partial support
The key differentiator. Write as Delta, automatically expose as Iceberg. No ETL, no data duplication.
UniForm eliminates the format debate entirely. A single physical table is simultaneously readable as both Delta Lake and Apache Iceberg. UniForm metadata is generated as part of the Delta transaction, not as a separate batch process, so Iceberg readers always see a consistent view.
Query Delta Lake and Iceberg tables together in a single SQL statement
Delta Forge is not limited to one format at a time. Because both Delta Lake and Apache Iceberg are first-class citizens in the same SQL engine, you can join across formats in a single query with no ETL, no data movement, and no connectors to configure.
-- Join a Delta Lake table with an Iceberg table
SELECT d.customer_id, d.order_total, i.shipping_status
FROM delta.warehouse.orders d
JOIN iceberg.warehouse.shipments i
ON d.order_id = i.order_id
WHERE d.order_date >= '2024-01-01';Production-grade row-level deletes with the latest Iceberg capabilities
Row-level deletes are where many Iceberg implementations fall short. Delta Forge supports the full spectrum of delete mechanisms defined in the specification, from position deletes in V2 through bitmap-based deletion vectors in V3.
Bitmap
{1, 3, 6}
~12 bytes
Evolve your schema without rewriting data files
Add new columns at any position. Existing files return NULL for new columns automatically.
ALTER TABLE t ADD COLUMN new_col STRING
Remove columns from the schema. Underlying data remains until compaction.
ALTER TABLE t DROP COLUMN deprecated_col
Rename columns using Iceberg's field ID tracking. Zero data movement required.
ALTER TABLE t RENAME COLUMN old_name TO new_name
Change column order for better organization without touching data files.
ALTER TABLE t ALTER COLUMN col FIRST
Widen column types safely (int to long, float to double, decimal precision).
ALTER TABLE t ALTER COLUMN amount TYPE DECIMAL(20,4)
Evolve struct, list, and map types at any nesting depth. Add fields to nested structures.
ALTER TABLE t ADD COLUMN address STRUCT<zip: STRING, city: STRING>
Query any historical state of your Iceberg tables
SELECT * FROM iceberg.warehouse.events VERSION AS OF 5
SELECT * FROM iceberg.warehouse.events TIMESTAMP AS OF '2024-06-15'
SELECT * FROM events.branch_staging
CALL rollback_to_snapshot('events', 5)
Manifest-level and file-level statistics for efficient query planning
Iceberg's metadata hierarchy enables aggressive data skipping at multiple levels. Delta Forge leverages partition pruning, manifest filtering, and file-level column statistics to minimize I/O and maximize throughput.
Native Iceberg and Delta Lake support in a single engine. Your data, your format, your choice.