An integration caches full UCS and Intersight compute inventory documents that differ by product family and are almost always read as whole objects after each API revision. Which store type is the best fit?

A document store that tolerates schema-flexible, API-shaped JSON records is the best fit when the integration caches full UCS and Intersight compute inventory documents that differ by product family and are almost always read as whole objects after each API revision. Variable-shaped payloads behave like differently packed moving boxes: you usually want the whole box. Document databases embrace that flexibility without a migration for every additive field. Reading entire inventory documents matches the stated access pattern better than normalizing every nested attribute into rigid tables.

A rigid relational schema that must be altered on every API field change fights revision velocity and creates constant migration work for little join benefit when whole documents are the unit of read. A columnar warehouse whose strength is scanning two metrics across billions of rows targets analytic aggregates, not fetching complete compute inventory objects. A time-series database keyed only by metric name and timestamp stores measurements, not heterogeneous UCS or Intersight inventory documents.

Exam caveat: document flexibility still needs application-level validation and versioning so silent field drift does not break consumers; schemaless is not the same as no contract. Operational check: ingest two product-family inventory documents with differing shapes into the cache store, read each back as a whole object after a simulated API revision that adds a field, and confirm no schema migration was required for the read path.