Table Storage

Azure Table Storage is a cheap, schemaless key-value and wide-column store for large volumes of structured, non-relational data. It lives inside a storage account, costs a fraction of a cent per gigabyte, and serves simple lookups by key at scale without any of a database's management.

It is deliberately minimal: no secondary indexes, no joins, no complex queries, no global distribution. When an application needs those, the upgrade path is the Cosmos DB Table API, which speaks the same protocol with far more capability — at a far higher price. Table Storage is the right answer precisely when the access pattern is simple and the budget matters.

Data Model

Each entity has a partition key and a row key, and together they form the primary key. The partition key groups entities for scalability and transactions; the row key uniquely identifies an entity within a partition. Beyond those, properties are schemaless — different entities in the same table can carry different fields. Designing for this two-part key is the whole art of Table Storage.

Performance and Partitioning

Lookups by partition key and row key are fast and scale well; queries that scan across partitions or filter on non-key properties are slow and grow worse with data volume. A well-chosen partition key spreads load and keeps related entities together; a poor one creates hot partitions or forces table scans. As with Cosmos, the key design decides whether it is fast or unusable.

When to Use It

Table Storage fits telemetry, logs, device data, and user or session records with a known, key-based access pattern and a tight budget. It does not fit anything needing rich queries, secondary indexes, transactions across partitions, or low-latency global reads — those are Cosmos DB's job, and reaching for Table Storage there leads to slow scans and workarounds.

Limits

Table Storage has no secondary indexes, no server-side joins, and limited transaction scope (entity-group transactions within a single partition only). It inherits its storage account's redundancy — including geo-redundant GRS and GZRS for cross-region durability — but that is async DR replication, not active global distribution. The Cosmos DB Table API offers the same API surface with automatic indexing, true multi-region distribution with local reads and writes, and guaranteed throughput — the path when an application outgrows these limits.