What it is

Sequin is an open-source change data capture platform for Postgres. It lives in the Postgres ecosystem and reads database changes, then sends those changes to streams, queues, search indexes, and other endpoints. The project is written in Elixir and released under the MIT license.

The concrete problem it solves is moving Postgres data changes out of the database without building separate pipelines for each destination. It can backfill existing rows and stream new changes in real time. This lets a team send the same change stream to systems such as Kafka, GCP Pub/Sub, Elasticsearch, Meilisearch, NATS, RabbitMQ, Kinesis, Azure EventHubs, HTTP endpoints, or another Postgres database.

Key capabilities

• It captures Postgres change data and turns row changes into events that other services can consume.
• It can backfill existing rows and then continue streaming new changes in real time.
• It publishes to Apache Kafka, Amazon Kinesis, Azure EventHubs, GCP Pub/Sub, NATS, RabbitMQ, Elasticsearch, Meilisearch, and HTTP endpoints.
• It supports a Sequin Stream sink where consumers read changes directly with HTTP GET and exactly-once processing.
• It can keep search indexes in sync with Postgres for engines such as Elasticsearch, Meilisearch, and Typesense.
• It can maintain caches by streaming database changes to Redis compliant caches.
• It can create audit logs by tracking, enriching, and recording every change in the database.

Who uses it and how

• Teams use it to stream database changes as events for services that consume Kafka, SQS, Pub/Sub, or other queue systems.
• Compliance and platform teams use it to build audit logs from every change in a Postgres database.
• Application teams use it to keep caches in sync by sending Postgres changes to Redis compliant caches.
• Search teams use it to warm and refresh Elasticsearch, Meilisearch, or Typesense indexes with sub-second latency.
• Infrastructure teams deploy it as a standalone Docker container next to Postgres, and AWS teams can use Terraform templates from the deployment directory.

Getting started

The typical install path is a standalone Docker container placed next to the Postgres database. For production deployments on AWS, the README points to Terraform infrastructure-as-code templates in the deployment directory.

When to use it — and when not to

Sequin is useful when a team wants Postgres change data to reach many sinks from one open-source system, and when it can operate the Postgres, Docker, AWS, and destination services itself. The README compares it with Debezium, Fivetran, and Airbyte, and states throughput of 50,000 operations per second, 40 MB per second, 55 milliseconds average latency, and 253 milliseconds 99th percentile latency. It is less suitable when a team cannot run and maintain the container and its configuration.