ref-sp-db-nosqllisted

# NoSQL Databases ## Purpose Provide portable defaults for deciding when non-relational databases fit better than a purely relational model and for designing their tradeoffs consciously. ## When to use this skill - Choosing between relational and non-relational persistence for a new workload. - Designing document, key-value, graph, or wide-column data models. - Planning horizontally scaled or geographically distributed data systems. - Reviewing denormalized schemas, duplication strategy, or eventual-consistency behavior. - Deciding how a NoSQL store should coexist with a relational system of record. ## Scope boundaries This skill owns the **store-choice decision** — relational, non-relational, or hybrid — and non-relational modeling. Once the answer is "relational", hand off. - `ref-sp-db-schema-design` — relational modeling, keys, constraints, and normalization. - `ref-sp-db-operations` — transactions, indexing, recovery, and migration for whichever store wins. - `ref-sp-db-distributed` — site topology, fragmentation, and replica placement. Distribution is a separate axis from the data model: distributed databases are not automatically NoSQL, and NoSQL systems are not automatically distributed. ## Defaults - Choose the database model from access patterns, consistency needs, and operational constraints, not trend-following. - Treat denormalization as a deliberate modeling technique in NoSQL, not as a failure to normalize. - Keep stable identifiers so related data