Most lakehouse jobs don't need a
cluster. They need a native engine.
Read and write Delta and Iceberg in place. Serve BI directly from the lake, run graph workloads on the same data, and give agents typed catalog access.
All on your hardware. No second warehouse. No copy pipeline. No cluster tax.
Read and write Delta and Iceberg
with proofs you can run
Delta Lake and Apache Iceberg in one native engine, with first-class grammar for MERGE, time travel, deletion vectors, UniForm interop, and change data feed. Correctness is not claimed, it is checked: two independent verification layers, with every expected value derived outside the engine under test.
Pay only for the work
you actually run
DeltaForge is one native engine on hardware you already pay for. Compute is metered as core-seconds while a query runs: an idle node costs nothing, and a faster engine spends fewer of them.
Pay for queries, not standby
Compute is metered while a query actually runs. An idle node bills nothing: no per-row scan fees, no per-API charges, no minimum cluster uptime.
Scale with load
Stateless, quick-starting workers fit ordinary Kubernetes autoscaling: pods scale up when queries arrive and back down when they stop.
Faster is cheaper
5x to 8x faster than Spark on standard reads, ~4x on writes. The same answer costs fewer core-seconds. See the numbers.
No second warehouse
BI tools read the lake directly through ODBC and ADBC. No duplicate copy to move, license, govern, or keep in sync.
Honest numbers,
same tables, same hardware
Four standardized read suites and one write workload, run against the same plain-Delta fixtures by DeltaForge, DuckDB, and two Spark profiles. DuckDB wins reads by 1.5x to 2.5x; DeltaForge beats both Spark profiles by 5x to 8x on every suite. The whole harness is Apache 2.0, reproducible in one command, and every tie or loss is reported by name.
| Benchmark | DeltaForge | DuckDB | Spark default | Spark tuned | Detail |
|---|---|---|---|---|---|
| TPC-H 22 queries, 8 tables | 255 ms | 173 ms | 1,478 ms | 1,528 ms | tpch.md |
| TPC-DS 99 queries, 24 tables | 271 ms | 171 ms | 1,568 ms | 1,464 ms | tpcds.md |
| SSB 13 queries, 5-table star | 191 ms | 75 ms | 685 ms | 628 ms | ssb.md |
| JOB 113 queries, IMDB | 976 ms | 632 ms | crashed* | crashed | job.md |
Warm-median ms per suite at SF=1, lower is better. *Spark default crashed after q06d on JOB and Spark tuned failed to start; no median is published for partial runs. DeltaForge and DuckDB completed all 113 queries.
Writes: ~4x faster than Spark on single-node Delta
10,000,000-row CTAS into plain Delta from a deterministic synthetic source, same nine-column schema and row content. DeltaForge: 6.48M rows/sec · Spark default: 1.51M · Spark tuned: 1.61M. DuckDB sits this one out because its delta extension is read-only. Full write benchmark.
Power BI, Tableau, Excel
directly on your lake
BI usually means copying the lake into a second SQL database first. DeltaForge ships two drivers that point at Delta directly: a native ODBC driver for the whole ODBC ecosystem, and an ADBC driver with a bundled Power Query connector for Power BI Desktop. Same engine, same governance, no copy.
Your Delta tables
become native property graphs
Fraud rings, supply chains, and customer-to-merchant patterns usually mean a separate graph database and a sync pipeline to feed it. DeltaForge projects your existing Delta tables into a native property graph in place: traverse it with Cypher, score it with PageRank, and join the results back to SQL in the same session.
Typed catalog access
for code assistants and agents
Assistants and automation need typed access to the real catalog, not guesses from a prompt. DeltaForge ships a built-in MCP server: plug it in once and Claude, Cursor, or Copilot get typed actions for catalog, lineage, SQL, and pipelines, under the same RBAC and audit logging as a human user.
Four real-world workloads
one native engine
Heavy Delta and Iceberg SQL with proof it works. BI served straight from the lake. Cypher and graph algorithms over the same tables. Agents wired into the catalog through MCP. One native engine, on your infrastructure.
