QueenMQ talks to the systems around it from inside the broker: Kafka clients through a Kafka port, S3 buckets through a sink, PostgreSQL tables through a source and a sink. There is no extra process to run. Each one is a few settings or one HTTP call, and each one runs on every node of a cluster.
| Integration | What it does |
|---|---|
| Kafka clients | Kafka producers, consumers and admin tools use QueenMQ unchanged: a topic is a queue, a consumer group is a group. |
| Kafka frameworks | Kafka Streams, Kafka Connect and Flink against the Kafka port, with the settings each one needs. |
| S3 sink | The messages of chosen queues, written to a bucket as JSONL or Parquet, once each. |
| PostgreSQL | A table streamed into a queue through logical replication, and a queue written into a table, with exactly-once effects. |
| Laravel | A queue driver for Laravel: your jobs stay Laravel jobs, with order per customer and a supervisor that follows the backlog. |
Why
every process next to a broker is one more thing to deploy, watch and upgrade. Inside the broker, an integration shares the broker’s cluster, its upgrades and its guarantees.