Design a database sharding strategy for a multi-tenant SaaS platform experiencing rapid write growth. The schema includes large relational tables that require cross-shard joins for reporting. Detail your approach for: 1) Choosing a sharding key (tenant ID vs. hash vs. range) and the trade-offs involved. 2) Handling 'hotspots' where specific tenants generate disproportionate load. 3) Implementing a fan-out query mechanism or a data aggregation layer to handle reporting across shards without degrading performance. 4) Strategies for re-balancing data across shards with minimal downtime. Discuss the consistency guarantees your approach provides.