Design a high-throughput, distributed message queuing system capable of handling millions of events per second with persistent storage, partition-level ordering guarantees, and support for multiple consumer groups. Ensure the design is fault-tolerant and horizontally scalable.