CQRS separates the two responsibilities that a single model usually conflates: commands (writes that change state) and queries (reads). The write model is normalized, validated, and consistent — optimized for correctly enforcing business rules and invariants (Phase 6). The read model is denormalized and purpose-built for fast queries — often a materialized view or summary table that answers a query in a single fast read with no joins (exactly the denormalized read models from Phase 6). The two are typically kept in sync via events (Phase 8), which means the read model may lag slightly behind the write model — eventual consistency, a deliberate tradeoff. The payoff is that reads and writes are optimized independently: the read side can be denormalized and cached (Phase 7) for a query-heavy workload without complicating the write side's consistency logic. This directly builds on the denormalization and read-model concepts you learned in Phase 6.