I am not the creator of this library, but the choice of Redis makes sense.

• MongoDB doesn't support atomic operations. Redis does. This is important for dealing with concurrency.
• MySQL doesn't provide good support for structured JSON data before version 8.
• Generally, hosting Redis tends to be less resource intensive and therefore cheaper than MySQL (and probably MongoDB).
• Neither MongoDB nor MySQL support pub/sub which is used to report results back on jobs run.
• In-memory means it runs faster than a transactional database would typically run, reducing the bottleneck impact of bee-queue on an application overall.

See https://web.archive.org/web/20220504052812/https://github.com/bee-queue/bee-queue/wiki/Origin for what the original authors were thinking when they wrote this library.