Azure Table storage

What it is

Azure Table storage is a NoSQL key-value / key-attribute store for structured, semi-structured data — billions of rows without a fixed schema. Each row (entity) has a partition key + row key (the composite primary key) and attribute columns.

Why it exists

Cost-effective, schema-less storage for structured app data (user profiles, device metadata, logs) that doesn’t need a full relational database. Tables scale out by partition for parallel throughput.

Key ideas

  • Entity = row; attributes = columns; a partition groups entities by partition key for atomic operations & load distribution.
  • Composite primary key: PartitionKey + RowKey (together unique).
  • Schemaless — entities can have different attributes; via the OData/REST table service.
  • Cheap per-GB — good for large structured datasets with simple lookups.
  • Part of a storage account alongside blobs/queues/files; addressable at <account>.table.core.windows.net.
  • Managed offering Azure Cosmos DB (Table API / Azure Table) can replace it for premium.

How it fits (diagram)

azure-tables.svg

Exam notes

  • PartitionKey + RowKey = primary key; partition = unit of scale & atomic batch.
  • Tables are schema-less — use for structured, not relational.
  • Contrast with queues (messages, transient) vs blobs (unstructured files).

storage-account · azure-queues · blob-storage

📘 Source: Microsoft Learn — Azure Tables