LEARN. PRACTICE. GROW.Concept and mechanism
Blob Storage keeps objects such as documents and recordings accessible through APIs. Azure Files provides shares using supported protocols such as SMB or NFS, useful when applications require file operations. Table storage keeps NoSQL entities with partition and row keys; Table in its name does not imply relational joins and foreign keys. Data Lake Storage adds capabilities such as hierarchical namespace for analytical file patterns. Selection depends on the access contract rather than just file extension. An application mounting SMB does not automatically work over an object container without adaptation.
Guided application
Cosmos DB provides API families for different models and clients, including NoSQL, MongoDB, Cassandra, Gremlin, and Table. Do not assume interchangeable queries and drivers. Current documentation also distinguishes Azure DocumentDB vCore; a family name does not settle concrete product compatibility. In a fictional request registry, choose the partition key around distribution and queries. If almost every write uses country=PT, buying global throughput may not remove a hot partition. Measure the pattern, assess cost, and prepare any key migration in a controlled way. Handover includes API, identity, partitioning, retention, and throttling diagnosis, avoiding universal promises about scale or consistency.
Renaming the key from PT to Portugal retains the same load concentration.
Common pitfalls
Object as share; Table as SQL; API as a mere name; total throughput as guaranteed distribution.
Related topics: Data, workloads, and responsibilities · Relational modeling, integrity, and SQL · Relational services and compatibility
Choose around the access pattern and measure actual load distribution.
Reference: Data store models and access patterns · DP-900 skills measured 2026-07-21