Other databases

Status: Exploration · Priority: 6

What

Extend Brume’s support beyond PostgreSQL to:

• MySQL (and MariaDB)
• SQL Server
• MongoDB

Why it matters

PostgreSQL is Brume’s first-class home, but many teams operate polyglot data stacks. Asking them to maintain a different pseudonymization tool per engine fragments their compliance story and multiplies the secrets they have to protect.

What it’ll unlock

• A single pseudonymization tool across heterogeneous stacks.
• Shared BRUME_HMAC_SECRET / BRUME_FPE_KEY across engines, with cross-engine linked_columns keeping the same fake value coherent.
• A consolidated DPO sign-off across the whole data platform, not just one engine.

Order of attack (tentative)

1. MySQL / MariaDB — closest cousin to Postgres, similar FK and constraint model, reuses most of the pipeline.
2. SQL Server — more enterprise demand, requires careful handling of schema and identity columns.
3. MongoDB — different paradigm (document store, no FKs). Needs a different abstraction layer; linked_columns becomes the primary mechanism.

Open design questions

• One unified config schema vs. per-engine YAML.
• How to map FPE on engines without a native equivalent to Postgres’s domain types.
• Schema discovery on MongoDB — strict schema vs. sampled.