Integrations

Nussknacker allows to reach out to the data and services provided via streaming and queuing systems, consumable computer services, databases, ML deployment platforms, data formats and catalogues. Integration definition allows to define shared integration parameters only once for multiple components. You can see what kind of components will be generated on the integration tile in the Integrations view. Nussknacker will automatically generate components relevant to the given integration, for example source, enricher or sink and will add them to the Component Palette in Designer.

Configuring a new integration

Open Integrations from the Designer's side menu (the view lives at the /configurator path), find the integration tile and click Add new. Each tile lists the kinds of components the integration generates (Source, Sink, Enricher).

note

The Integrations view is provided by the Integrations (configurator) Designer extension. It is available in Nussknacker Cloud and in Enterprise deployments, including the Docker Compose installation bundle. If the side menu shows no Integrations item (for example in the Open Source edition), configure integrations in configuration files - see the Open Source version integrations documentation.

Supported integrations

Message brokers

The following queueing and streaming systems are currently supported:

Integration	Type(s)
Aiven Kafka	Source / Sink
Azure Event Hubs	Source / Sink
Confluent Cloud	Source / Sink
(generic) Kafka	Source / Sink
MQTT	Source / Sink
Redpanda	Source / Sink
Table (Flink Table API)*	Source / Sink

*Check Flink Table API section for more details.

Databases

Integrations with database systems are implemented using multiple and diverse mechanisms, which results in different types of components being available. The table below summarizes different integration mechanisms and component types that they yield.

Integration	Components	Comments
plain JDBC	Enricher	No need to provide data type mappings between target database and data types used by Nussknacker
Flink connector available (Table API is used)	Source / Sink	The integration definition includes data type mappings between the database and Flink
Flink catalog support available	Enricher / Source / Sink	No need to provide data type mappings

The following database systems are supported:

Integration	Type(s)
Clickhouse	Source / Sink
MariaDB	Enricher / Source
MySQL	Enricher / Source / Sink
MS SQL Server (experimental)	Enricher / Source
Oracle	Enricher / Source
PostgreSQL	Enricher / Source / Sink
Redis	Enricher (Key-Value Store)
Redshift	Enricher
Snowflake	Enricher
Table (Flink Table API)*	Source / Sink

*Check Flink Table API section for more details.

MS SQL Server is marked as experimental in the integrations catalog.

The Source listed for MariaDB, MS SQL Server and Oracle is the SQL source; these databases have no Table source or sink.

Services:

Integration	Type(s)
OpenAPI	Enricher

ML Models Registries:

Integration	Type(s)
Databricks Unity Catalog	Enricher
MLflow	Enricher

Filesystems

Integration	Type(s)
Table (Flink Table API)*	Source / Sink

*Check Flink Table API section for more details.

Data lakehouse (Apache Iceberg)

Integration	Type(s)
AWS Glue	Source / Sink

See AWS Glue (Iceberg REST catalog) for details.

Flink Table API

The Flink's Table API provides a relational abstraction for unified stream and batch processing, allowing to define logical tables in Flink that map to data stored in external systems such as databases, message queues or file systems using SQL-like syntax. See Table API integration for more details.

AI Services

AI Providers

Integration	Type(s)
Amazon Bedrock	Enricher
Anthropic	Enricher
Google AI Gemini	Enricher
MistralAI	Enricher
Nvidia	Enricher
OpenAI	Enricher
OpenAI-compatible	Enricher
Decision AI (Jev)	Decision Model enricher

Vector databases

Integration	Type(s)
Qdrant	Enricher
Weaviate	Enricher

Miscalaneous

Integration	Type(s)
MCP Server	N/A

Integration missing?

If the provider or system of your choice is not supported yet, please contact us via info@nussknacker.io - we will handle the data types mappings and other technical details for you.