DP-900 · 3 Describe considerations for working with non-relational data on Azure
Azure Table storage
Exam objective: Describe features of Azure Table storage
Azure Table storage is a NoSQL key value store where every row has a partition key and a row key, but rows in the same table can hold different columns.
Azure Table storage is a NoSQL service that stores rows inside tables, but those tables do not work like a relational table. Every row needs a unique key made of two parts, a partition key and a row key, and Azure records a timestamp automatically whenever a row changes. Beyond those columns, each row is free to carry a different set of fields.
Because of that, data in a table is usually denormalized. A customer row might hold the name, several phone numbers and several addresses all in one row, fields that a relational design would split across multiple related tables instead.
The partition key is central to how Table storage performs well. Rows that share a partition key are stored together, and a table can contain any number of partitions. When a query includes the partition key, Azure can go straight to the right partition instead of scanning the whole table, which speeds up a point query for one row and a range query for a contiguous block of rows.
On the exam, look for partition key, row key or key value data.
Key points
- Table storage holds semi-structured data: every row carries a partition key, a row key and a timestamp, while the remaining columns can vary between rows.
- Data is usually denormalized, with one row holding everything about a logical entity instead of spreading it across several related tables.
- A table has no foreign keys, relationships, stored procedures or views, unlike a relational database.
- The partition key groups related rows together, and partitions can grow or shrink independently of each other.
- Including the partition key in a query narrows the search and reduces the reads needed to find a row.
Exam trap
The name is misleading. An Azure table is not a relational table, rows share only the key and timestamp columns, and every other column can differ from row to row.
Check yourself
A company stores customer profiles. One customer has three phone numbers and two addresses, while another has one phone number and no address, and all of this lives in a single row per customer. Which description fits this design?
Go deeper on Microsoft Learn
Checked against Microsoft Learn on October 1, 2026.