// the find
sequinstream/sequin
Postgres change data capture to streams, queues, and search indexes like Kafka, SQS, Elasticsearch, HTTP endpoints, and more
Sequin is a standalone CDC tool that reads Postgres WAL and delivers changes to a wide range of sinks — Kafka, SQS, Redis, Elasticsearch, webhooks, and more — without requiring you to run Kafka just to use it. It's for teams that want real-time Postgres CDC without the Debezium + Kafka infrastructure tax. Written in Elixir/OTP, which is the right choice for this kind of concurrent, fault-tolerant message routing.
The performance numbers are credible and impressive — 50k ops/s at 55ms average latency against Debezium's 258ms is not a rounding error, it's architectural. The sink list is genuinely broad: covering both streaming (Kafka, Kinesis, Pub/Sub) and direct targets (Redis strings, Elasticsearch, webhooks) means you don't need a second hop for simple use cases. Backfill support is a first-class feature, not an afterthought — being able to replay specific rows to a sink without re-running migrations is something Debezium handles poorly. YAML-based config with a Management API means you can version-control your CDC setup, which most competitors treat as a UI-only concern.
It's a relatively young project (2k stars, 138 forks) taking on a space dominated by Debezium, which has years of production hardening across pathological Postgres configurations — WAL gap handling, replication slot pressure under high write load, and schema evolution edge cases are where new CDC tools tend to break. The Elixir transform functions are a double-edged sword: low-latency but it means your ops team now needs to know Elixir to debug why messages are getting dropped. No mention of schema registry integration, which matters the moment your Kafka consumers need Avro or Protobuf contracts. The self-hosted deployment story is a single Docker container, which is fine for dev but the production Terraform templates in `deployment/` suggest AWS-only IaC — Azure and GCP deployments are left as an exercise.