Skip to main content



The Airbyte Kafka destination allows you to sync data to Kafka. Each stream is written to the corresponding Kafka topic.


  • For Airbyte Open Source users using the Postgres source connector, upgrade your Airbyte platform to version v0.40.0-alpha or newer and upgrade your Kafka connector to version 0.1.10 or newer

Sync overview

Output schema

Each stream will be output into a Kafka topic.

Currently, this connector only writes data with JSON format. More formats (e.g. Apache Avro) will be supported in the future.

Each record will contain in its key the uuid assigned by Airbyte, and in the value these 3 fields:

  • _airbyte_ab_id: a uuid assigned by Airbyte to each event that is processed.
  • _airbyte_emitted_at: a timestamp representing when the event was pulled from the data source.
  • _airbyte_data: a json blob representing with the event data.
  • _airbyte_stream: the name of each record's stream.


Full Refresh SyncNo
Incremental - Append SyncYes
Incremental - Append + DedupedNo

Getting started


To use the Kafka destination, you'll need:

  • A Kafka cluster 1.0 or above.

Setup guide

Network Access

Make sure your Kafka brokers can be accessed by Airbyte.


Airbyte should be allowed to write messages into topics, and these topics should be created before writing into Kafka or, at least, enable the configuration in the brokers auto.create.topics.enable (which is not recommended for production environments).

Note that if you choose to use dynamic topic names, you will probably need to enable auto.create.topics.enable to avoid your connection failing if there was an update to the source connector's schema. Otherwise a hardcoded topic name may be best.

Target topics

You can determine the topics to which messages are written via the topic_pattern configuration parameter. Messages can be written to either a hardcoded, pre-defined topic, or dynamically written to different topics based on the namespace or stream they came from.

To write all messages to a single hardcoded topic, enter its name in the topic_pattern field e.g: setting topic_pattern to my-topic-name will write all messages from all streams and namespaces to that topic.

To define the output topics dynamically, you can leverage the {namespace} and {stream} pattern variables, which cause messages to be written to different topics based on the values present when producing the records. For example, setting the topic_pattern parameter to airbyte_syncs/{namespace}/{stream} means that messages from namespace n1 and stream s1 will get written to the topic airbyte_syncs/n1/s1, and messages from s2 to airbyte_syncs/n1/s2 etc.

If you define output topic dynamically, you might want to enable auto.create.topics.enable to avoid your connection failing if there was an update to the source connector's schema. Otherwise, you'll need to manually create topics in Kafka as they are added/updated in the source, which is the recommended option for production environments.

NOTICE: a naming convention transformation will be applied to the target topic name using the StandardNameTransformer so that some special characters will be replaced.

Setup the Kafka destination in Airbyte

You should now have all the requirements needed to configure Kafka as a destination in the UI. You can configure the following parameters on the Kafka destination (though many of these are optional or have default values):

  • Bootstrap servers
  • Topic pattern
  • Test topic
  • Sync producer
  • Security protocol
  • SASL JAAS config
  • SASL mechanism
  • Client ID
  • ACKs
  • Enable idempotence
  • Compression type
  • Batch size
  • Linger ms
  • Max in flight requests per connection
  • Client DNS lookup
  • Buffer memory
  • Max request size
  • Retries
  • Socket connection setup timeout
  • Socket connection setup max timeout
  • Max block ms
  • Request timeout
  • Delivery timeout
  • Send buffer bytes
  • Receive buffer bytes

More info about this can be found in the Kafka producer configs documentation site.

NOTE: Some configurations for SSL are not available yet.


Expand to review
VersionDatePull RequestSubject
0.1.102022-08-0415287Update Kafka destination to use outputRecordCollector to properly store state
0.1.92022-06-1713864Updated stacktrace format for any trace message errors
0.1.72022-04-1912134Add PLAIN Auth
0.1.62022-02-1510186Add SCRAM-SHA-512 Auth
0.1.52022-02-1410256Add -XX:+ExitOnOutOfMemoryError JVM option
0.1.42022-01-319905Fix SASL config read issue
0.1.32021-12-308809Update connector fields title/description
0.1.22021-09-146040Change spec.json and config parser
0.1.12021-07-305125Enable additionalPropertities in spec.json
0.1.02021-07-213746Initial Release