ANASy

Analytics Sync. Keep OLTP thin: publish fat events once to Kafka. Any number of sinks consume them.

OLTP  --EventPublisher-->  Kafka  --group A-->  ClickHouse
                           └────  --group B-->  other warehouse

Kafka is the product boundary. ClickHouse is a reference sink, not the architecture.

Modules

Module	Layer
event-connector-starter	Core. Inject EventPublisher.
sinks/clickhouse-batch-sink	Optional. Batch JDBC into ClickHouse.

Java 21, Spring Boot 3.4, Maven. Group io.anasy.

Docs

Doc	Read when
docs/PLAN.md	Locked decisions
docs/hld.md	Architecture, fan-out
docs/lld.md	Starter, publisher, how to add a sink
docs/scaling.md	Partitions, groups, per-sink tuning
docs/sinks/clickhouse.md	ClickHouse config, schema, queries
AGENTS.md	Agent rules

Quick path

1. Run Kafka. Depend on event-connector-starter.
2. Publish:

events.publish("orders.events", order.id().toString(), fatEvent);

3. Add a sink: new app, unique group-id, write record.key() as eventId. ClickHouse: apply docs/sinks/clickhouse.sql and run clickhouse-batch-sink.