Developers frustrated with Redis as a job queue backend have a new alternative: flashQ. This Rust-based job queue server promises to outperform traditional Redis-based solutions, offering 2,127,660 ops/sec compared to 36,232 ops/sec with BullMQ. The creator, dissatisfied with the network overhead and polling inefficiencies of Redis, developed flashQ to run jobs in-process, significantly reducing latency and improving throughput.

flashQ leverages Rust's strengths, including memory safety and fearless concurrency, to deliver a robust solution. It features 32 parallel shards, each running on a separate CPU core, enabling atomic batch operations. This architecture results in a 600x speed improvement for batch operations compared to Redis. Additionally, flashQ offers a range of features expected in a production-ready queue, such as priority queues, delayed jobs, and job dependencies.

The performance benefits are substantial. On an Apple M2, flashQ achieved a 58x improvement in batch throughput and a 52x increase in pull and acknowledge operations. Latency is also dramatically reduced, with P99 latencies ranging from 127-196 μs, compared to 606-647 μs with BullMQ. Memory efficiency is another highlight, with flashQ using only 200 MB per 1M jobs, a 10x reduction from BullMQ's 2 GB.

flashQ is designed to be a drop-in replacement for BullMQ, maintaining compatibility with existing APIs. This allows developers to switch with minimal code changes, simply by altering the import statement. The project is open source, licensed under MIT, and actively developed, with the current version being v0.2. Developers are encouraged to try it out and report any issues, as the creator seeks community feedback to further refine and enhance the tool.