Stackbook Logo
architecture-descriptionestablished · high operational burden

CQRS (Command Query Responsibility Segregation) Patterns

Also known as: cqrs, command-query-separation, read-write-separation

Intent

Separate read and write models to optimize each: writes for consistency, reads for query performance and flexibility.

Problem

Single model serves writes (invariants, normalization) and reads (joins, denormalization) — compromises both.

Forces

  • •Write model needs: transactions, invariants, normalization
  • •Read model needs: denormalization, pre-computed joins, caching
  • •Read/write scaling differs (reads >> writes typically)
  • •Complex queries on normalized write model are slow

Solution

✓ When to Use

  • •Read/write patterns differ significantly
  • •Complex queries on normalized data
  • •Different scaling needs (read replicas vs write primary)
  • •Multiple read representations (UI, reporting, API)

✗ When Not to Use

  • •Simple CRUD with straightforward queries
  • •Strong consistency required between read/write
  • •Team not ready for eventual consistency
  • •Overhead not justified by query complexity

Pros

  • +Each model optimized for its job
  • +Independent scaling of read/write
  • +Read model evolution without write impact
  • +Natural fit with event sourcing

Cons

  • −Eventual consistency between write and read
  • −Complexity: two models, synchronization, projection lag
  • −Duplicated logic risk (validation in both?)
  • −Operational burden: monitor projection health

Cost Profile

Infrastructure

Medium — separate read/write stores, projection workers

Operational

High — monitor lag, handle projection failures

Cognitive

Medium — model separation, consistency boundaries

Failure Modes

  • Projection lag → stale reads (user sees old data)

  • Projection failure → read model corrupt, needs rebuild

  • Write model invariants not enforced in read model

  • Dual-write inconsistency if not using event sourcing

Real-World Examples

Alternatives

  • traditional-layered
  • event-sourcing
  • materialized-views
  • read-replicas

Related Patterns

  • event-sourcing
  • materialized-view
  • read-replica
  • outbox-pattern
  • event-carried-state-transfer

Competency Domains

data statedistribution communicationreliability opseconomics evolutionscaling