Operational practice — database environments.\n\n"Most companies keep at least two versions of any database they are using."

Introduction / Context:
Organizations commonly operate multiple environments (development, test, staging, production) and rely on backups, replicas, and snapshots. However, saying they keep “at least two versions of any database” is a vague and misleading generalization that confuses versions with environments, replicas, or releases.

Given Data / Assumptions:

• “Version” could refer to schema version, software release, or live copies.
• Companies may have many non-production copies, but that does not imply two production “versions” concurrently.
• Replication copies are not distinct “versions”; they are synchronized instances.

Concept / Approach:
Best practices include maintaining separate environments and backups. Yet a single authoritative production database is the norm. Multiple schema versions may exist during rolling deployments, but they are transitional. Thus, the blanket claim that “most companies keep at least two versions” is not reliably true and muddles concepts.

Step-by-Step Solution:

Verification / Alternative check:
Examine deployment runbooks: migrations move a single database through versions; only one version is live per environment, with rollbacks if needed.

Why Other Options Are Wrong:

• Claims of “always” or tying to specific technologies do not match industry practice.
• Replication lag does not create a distinct version by design intent.

Common Pitfalls:
Equating read replicas or DR sites with different versions; assuming data warehouses and OLTP databases are “versions” of each other.

Final Answer:
Invalid (the statement overgeneralizes; practices vary)