# AutoMQ

> Kafka with its storage layer rebuilt on object storage

AutoMQ keeps Apache Kafka's compute layer and replaces its storage with cloud object storage, claiming 100% Kafka API compatibility and substantially lower cost by removing multi-replica local disks and cross-zone transfer.

Website: https://www.automq.com

## Use it when

- You run Kafka in the cloud and the bill, not the protocol, is the problem: diskless brokers over object storage remove replica storage and cross-zone traffic.
- Everything must stay Kafka-compatible; clients, Connect, and Streams work unchanged against its 100% Kafka API surface.
- You want elasticity without pain; brokers hold no persistent data, so partition reassignment takes seconds and scaling copies nothing.
- You want the option open source keeps: Apache-2.0 and self-hostable, unlike proprietary diskless rivals such as WarpStream.

## Think twice when

- Your latency budget is tight; writing through object storage trades some end-to-end delay against local NVMe Kafka.
- You run on-premises without an S3-class store; the design assumes cloud object storage underneath.
- The deployment is small and stable; plain Kafka or Redpanda may be simpler than adopting a younger rebuild.

## How it runs

Kafka brokers and KRaft controllers rebuilt over object storage with a write-ahead log in front; brokers keep no persistent local data. Self-host it free under Apache-2.0, or use AutoMQ Cloud billed through cloud marketplaces. Table Topic can write Iceberg tables directly from the stream.

## Details

### General

| Attribute | Value |
| --- | --- |
| License | Open source (permissive) |
| SPDX identifier | Apache-2.0 |
| Deployment | Self-hosted, Managed, Hybrid |
| Workload | Streaming |
| Operational complexity | Medium |
| Pricing | Free, Subscription |
| Language | Java |
| Repository | https://github.com/AutoMQ/automq |
| Documentation | https://docs.automq.com |
| Stars | 10.8k |

### Event Streaming

| Attribute | Value |
| --- | --- |
| Protocol | Kafka: 100% API compatible |
| Consumption model | Replayable log; consumers track offsets |
| Storage | Object storage; brokers hold no persistent local data |
| What you run | Kafka brokers and KRaft controllers over object storage, plus a WAL |
| Scaling | Reassignment in seconds; brokers scale without copying partitions |
| Multi-region | One logical cluster across regions; synchronous S3 writes, RPO=0 |
| Delivery guarantee | Kafka's: Exactly-once with transactions |
| Connectors | Kafka Connect unchanged; Table Topic writes Iceberg directly |
| Transformations | Kafka's own: Streams and Connect SMTs |
| Billing unit | Free to self-host; cloud usage billed through the marketplace |

## Capabilities

- [Event Streaming](https://matca.io/capabilities/event-streaming)

---

Source: https://matca.io/automq
Last updated: 2026-08-31T11:40:48.747Z