database-engine-selection-and-migrationlisted

# Database Engine Selection and Migration ## Purpose Make the accepted trade-offs and non-portable assumptions visible before an engine choice or cutover. Start with vetoes and behavioral proof, not a weighted feature popularity score. ## Decision inputs ```text read/write/admin SLOs, availability, RPO/RTO, retention, and growth: OLTP/analytic/hybrid workload, data shape/distribution, working set, and peak concurrency: transaction invariants and anomalies the domain permits or forbids: topology, regions, replication/CDC, backup/restore, and failover requirements: required SQL/types/extensions/indexes/search/JSON/partitioning/columnar capabilities: JDK, ORM, pool, exact driver versions, batching/fetch/generated-key behavior: edition/license/managed-service constraints and team operational competence: DDL/cutover window, reversibility, data sovereignty, and exit cost: ``` If these are missing, produce the evidence plan rather than selecting a winner. Inspect build/toolchain, resolved dependencies and runtime images for the actual JVM stack; this skill declares no universal JDK/driver baseline. A recommendation does not authorize an upgrade. Scope the evidence plan to missing decisive inputs, preserving conclusions already supported by the available evidence. ## Greenfield workflow 1. Express hard vetoes as testable requirements. A required behavior that an allowed edition or topology cannot provide eliminates an option; many small advantages do not cancel a veto. 2. W