LavinMQ: A High-Performance, Open-Source Message Broker Built in Crystal

Message brokers are a critical component of modern distributed systems, enabling asynchronous communication and decoupling services. However, traditional brokers can sometimes be resource-intensive or present operational challenges that are difficult to diagnose and resolve. LavinMQ emerges as a compelling open-source alternative, designed from the ground up by the experienced team behind CloudAMQP to address these very issues, promising exceptional performance and resource efficiency.

This article delves into LavinMQ's origins, its unique architectural choices, and the impressive benchmarks it has achieved, offering insights into how a message broker built with the Crystal language can redefine expectations for throughput and operational simplicity.

The Genesis of LavinMQ: Solving Real-World Broker Challenges

The creators of LavinMQ, the team at CloudAMQP, have spent 14 years hosting RabbitMQ, gaining deep insights into the operational complexities and pain points associated with message brokers. Carl Hoerberg, the creator of LavinMQ, explains that despite their extensive experience, customers occasionally encountered issues that were difficult to fully explain or work around within the constraints of an existing, external broker.

This led to the ambitious decision to build their own broker, granting them full control over the stack and enabling them to provide solutions where they previously couldn't. The journey began with an open-source AMQP proxy designed to handle short-lived connections, mitigating heavy CPU churn. What started as a protocol implementation evolved into a much larger challenge: building a reliable persistence layer. While the protocol itself proved to be the 'easy part,' achieving robust disk storage, acknowledgment handling, and replication took years of dedicated effort. LavinMQ's first release was in 2020, and it now powers over 5,000 production instances on CloudAMQP.

Technical Architecture and Performance Prowess

LavinMQ is engineered for speed and efficiency, distinguishing itself through several key design choices:

Language and Concurrency Model

LavinMQ is written in Crystal, a language known for its Ruby-like syntax combined with the performance benefits of an LLVM-compiled language and Go-like concurrency. This combination allows for highly efficient, type-safe code that compiles to native binaries, minimizing runtime overhead.

"Impressive benchmarks. Reaching 1M msgs/sec on a 2 vCPU instance is a great showcase for Crystal. As a Crystal core member, I've always seen LavinMQ as a prime example of performance-oriented engineering, especially with how it handles I/O and minimizes syscalls to get the most out of the hardware. If you want to see Crystal's concurrency and type system in a serious production environment, this is the project to check out. Kudos to the LavinMQ team for their work since 2020." — @sdogruyol, Crystal core member

Resource Optimization

The core design philosophy revolves around minimizing memory copies, allocations, and syscalls. Messages are written directly to disk via memory-mapped files, bypassing in-memory caches. Both messages and acknowledgments utilize append-only writes, a strategy that optimizes disk I/O and simplifies crash recovery.

Supported Protocols and Deployment

LavinMQ supports a range of essential messaging protocols, including AMQP 0-9-1, MQTT, HTTP, and streaming. It is distributed as a single binary, simplifying deployment. For quick evaluation, it can be run via Docker with a simple command: docker run -p 15672:15672 -p 5672:5672 cloudamqp/lavinmq.

Benchmark Highlights

The performance figures are a testament to its optimized design:

• ~580,000 messages/second on an AWS t4g.micro instance (2 vCPU, 1 GB RAM).
• Over 1,000,000 messages/second on an AWS c8g.large instance (2 vCPU, 4 GB RAM).

These numbers underscore LavinMQ's ability to deliver high throughput on minimal hardware, making it a compelling choice for cost-sensitive or high-performance environments.

Considerations for Reliability and Durability

For any message broker, reliability and message durability are paramount. A crucial question for users evaluating LavinMQ, particularly in distributed and fault-tolerant setups, concerns its behavior during leader failures. Specifically, whether a publisher confirm guarantees message survival on at least one follower in a clustered environment is a key aspect for ensuring data integrity and availability.

While the initial announcement highlights the challenges of reliable disk storage and replication, further details on how LavinMQ handles scenarios like leader failure in a clustered setup, particularly regarding publisher confirms, would be valuable for potential adopters. This is a common concern for systems requiring strong durability guarantees.

Conclusion

LavinMQ represents a significant effort to build a high-performance, resource-efficient, and operationally transparent message broker. Born from years of experience hosting RabbitMQ, it leverages the Crystal language and a meticulously optimized architecture to deliver impressive throughput on modest hardware. For organizations seeking an open-source message broker that offers a compelling blend of performance, efficiency, and control, LavinMQ presents a strong contender. The creators invite the community to explore their solution and share insights on current messaging practices and factors that would influence a switch to a new broker.