New · SOFI private deployment is ready for enterprise rolloutTalk to us→
SOFI
[ 45 SOURCES ][ 8 CATEGORIES ][ PUSHDOWN ][ ZERO COPY ]
endpoint · /integrations

Federate everyengine you own

45 sources across 8 categories — relational, cloud-managed, warehouses, NoSQL, streaming, files, APIs and SaaS. Every connector pushes down what it can; DuckDB federates the rest.

45

data sources

5

native relational engines

8

cloud-managed variants

8

data warehouses

4

output protocols

// connection typeNativeJDBCODBCS3-compatible
[ 01 / 03 ]Sources
// Connectors //

45 sources across 8 categories

Pushdown class shown per connector. Federation across mixed engines is stitched in DuckDB — full lineage preserved.

Relational

12 sources
PostgreSQL
GA
MySQL
GA
MariaDB
GA
SQL Server
GA
Oracle Database
GA
IBM Db2
GA
SAP HANA
GA
SAP ASE (Sybase)
GA
IBM Informix
Beta
Teradata
GA
H2
Beta
Firebird
Beta

Cloud & Managed

Connects via native engine protocol — Aurora/RDS as Postgres or MySQL, Cloud SQL as PG/MySQL/MSSQL

8 sources
Amazon Aurora· PG & MySQL
GA
Amazon RDS· all variants
GA
Azure SQL Database
GA
Azure Database for PG/MySQL
GA
GCP Cloud SQL· PG / MySQL / MSSQL
GA
Google AlloyDB
GA
Greenplum
GA
YugabyteDB / CockroachDB· PG wire-compatible
Beta

Data Warehouses

8 sources
Snowflake
GA
Google BigQuery
GA
Amazon Redshift
GA
Databricks SQL / Delta Lake
GA
Azure Synapse Analytics
GA
Microsoft Fabric Warehouse
GA
ClickHouse· self-hosted + Cloud
GA
Hive / Impala / Trino
GA

NoSQL, Graph & Search

6 sources
MongoDB· Atlas + self-hosted
GA
Amazon DocumentDB· via Mongo driver
GA
Cassandra / DataStax / ScyllaDB
GA
Elasticsearch / OpenSearch
GA
Neo4j· Cypher
Beta
HBase· via Phoenix
Beta

Streaming & Events

Consume topics as virtual tables with schema registry support

1 source
Apache Kafka· Confluent, MSK, Aiven
GA

Files & Object Storage

3 sources
Local files & shares· CSV, JSON, Parquet, Excel
GA
Amazon S3· CSV / JSON / Parquet
GA
MinIO / S3-compatible
GA

APIs & SaaS

4 sources
REST APIs· OpenAPI / Swagger discovery
GA
OData v4
GA
SOAP / Web Services· WSDL
GA
Salesforce· SOQL + REST
GA

Federation & Legacy

Migration, interoperability and generic connectors

3 sources
Denodo Virtual DataPort· peer federation / migration path
Beta
LDAP / Active Directory· as data source
GA
Any ODBC-capable source· generic ODBC connector
GA
// pushdownFullfilters, joins, aggregates & CTEs run at the sourcePartialengine-native pushdown where possible; DuckDB handles the restRead-onlySOFI fetches rows; all operators run in DuckDB
[ 02 / 03 ]Endpoints
// Output //

Four protocols, one governed view

Every published view is reachable across all four channels — same RBAC, same masking, same audit log.

:5556

SQL (pgwire)

PostgreSQL wire protocol. psql, JDBC, DBeaver, Metabase, dbt — any PG-compatible client connects natively.

:443

REST API

JSON over HTTPS. Per-view endpoints, signed responses, OpenAPI spec auto-generated.

:443

MCP for AI

Model Context Protocol. Every view is also a typed MCP tool with RBAC and audit per call.

:5556

JDBC / ODBC

Driver bundles for Java and ODBC tooling. Same governed surface as pgwire — different protocol.

[ 03 / 03 ]FAQ
// Integration FAQ //

What infra leads ask first

Plain answers about connection types, pushdown classes, custom connectors and tenant isolation.

Yes. Enterprise plans include a custom connector via the SDK — we've shipped connectors for SAP HANA, proprietary mainframe drivers and internal data APIs under NDA. Most take 1–2 weeks of joint engineering.

// connect anything

Wire your first sources in an afternoon.

Trial includes all native connectors + JDBC bundle. Custom connector and ODBC available on Enterprise.

Start free trialTalk to sales