**Physical data architecture** is the lowest level of abstraction within [[Data architecture|data architecture]], specifying how data is implemented, stored, and accessed within concrete technology environments. Where [[Conceptual data architecture|conceptual]] and [[Logical data architecture|logical data architecture]] address business meaning and platform-independent structure respectively, physical data architecture translates those designs into the specific constructs of a chosen [[Database management system|database management system]], [[Cloud computing|cloud]] platform, or storage infrastructure — including tables, indexes, partitions, file formats, and access controls. A physical data architecture encompasses the detailed implementation decisions that determine system performance, scalability, and operational characteristics. These include the selection of [[Database|database]] technologies — such as [[Relational database|relational]], [[NoSQL|NoSQL]], [[Column-oriented DBMS|columnar]], or [[In-memory database|in-memory]] systems — as well as storage organization strategies including [[Database index|indexing]], [[Partition (database)|partitioning]], [[Data compression|compression]], and [[Data replication|replication]]. Physical models specify [[Data type|data types]] as supported by the target platform, [[Stored procedure|stored procedures]], [[View (SQL)|views]], and [[Materialized view|materialized views]], and are closely tied to [[Query optimization|query optimization]] and [[Database performance|performance tuning]] practices. Physical data architecture is also concerned with [[Data security|data security]] and [[Regulatory compliance|regulatory compliance]] at the implementation level, governing [[Encryption|encryption at rest and in transit]], [[Access control|access control]], [[Data masking|data masking]], and [[Audit trail|audit logging]]. In modern [[Cloud computing|cloud]]-native and [[Hybrid cloud|hybrid]] environments, physical architecture decisions extend to [[Object storage|object storage]] configurations, [[Compute|compute]] and storage separation, and platform-specific services offered by providers such as [[Amazon Web Services|AWS]], [[Microsoft Azure|Azure]], and [[Google Cloud Platform|Google Cloud]]. The physical layer must balance the logical requirements established upstream against operational constraints including cost, latency, throughput, and maintainability.