Databases / Selector

Find a database worth testing.

Include your workload, preferred cloud, existing engine, and operating constraints. Build a shortlist with an explained starting point, a credible alternative, and the assumptions to verify.

Database selector01 / Build a shortlist

What are you building?

Choose the main workload. Refine the constraints to see a starting point, an alternative, and the tradeoffs.

Main workload
Cloud / platform preference

A preferred ecosystem, not a claim that every database is available as a native service there.

Deployment scale
Read consistency

Strict: authoritative reads of committed data. Balanced: fresh reads for critical decisions, stale reads elsewhere. Flexible: stale reads are acceptable. This is separate from transaction isolation.

Refine for your team & budget
Who will operate it?
Budget priority
Existing engine to preserve

Choose an engine if compatibility with an existing application matters.

Suggested starting point

PostgreSQL

Relational

Your core work changes related records and needs constraints, joins, and transactions in one database.

Watch out for

Choose transaction isolation deliberately. Read Committed is the default; an ACID database does not make every business workflow serializable.

Also consider: MySQL

Your team already operates it well and its relational features cover your access patterns.

What would change the choice?

Revisit this choice if one write region cannot meet measured latency or failure requirements.

More options in this category

Products to compare, not an additional ranking.

Browse all engines and services →
Assumptions & deployment notes

Transactional workload · Growing app · Strict read expectations · Managed service · Balance cost and time · No preference · No existing engine

  • Prefer a managed offering where available; confirm backup coverage, maintenance windows, export options, and your responsibilities.
  • Compare provider fees with engineering time using a representative workload and a growth scenario.
  • Measure growth and slow queries before adding databases, replicas, or shards.
  • You want authoritative reads of committed data. Define the read path, transaction isolation, and required behavior during failures; the product name alone cannot guarantee this.
Read the candidate’s documentation ↗

A shortlist to test against your workload. How recommendations work →

Continue your work

Take the next step.