NoSQLAdvanced
When is more than one database justified?
“What does polyglot persistence cost, and when is it worth it?”
What this tests
- Cost of additional stores
- Consistency boundaries
Answers by level
Read the beginner answer first and notice what is missing.
Each additional store is a consistency boundary, a backup, a monitoring surface and an on-call burden. It is worth it when one workload is measurably beyond what the primary handles — search relevance at scale, embeddings, high-volume time series — not because a different store is theoretically better for it.
The common shape is Postgres as the system of record with Redis, a search engine, or a vector store fed from it.
Green flags · Red flags
Strong green flag · Treats each extra store as a consistency boundary to justify.
Green flags
- Names the operational cost per store
- Feeds secondaries from a record of truth
- CDC for sync
Red flags
- "Best tool for each job" with no cost awareness
- Dual-writes without acknowledging drift
Follow-up questions
F1
How do you keep a search index in sync with Postgres?
Scenario
A team proposes Postgres + Mongo + Elasticsearch + Redis for a CRUD app with 10k users. Your take?