MaxCompute stores all data in tables — the input and output of every computing task on the platform. A table is a two-dimensional structure of rows and columns, where each row is a record and each column holds values of a single data type. If you've worked with relational databases, MaxCompute tables work the same way: use SQL to query and analyze data.
MaxCompute supports two table types: internal tables and external tables.
| Table type | Where data lives | Best for |
|---|---|---|
| Internal table | MaxCompute storage | Structured data that you process entirely within MaxCompute |
| External table | OSS or Tablestore (outside MaxCompute) | Unstructured data — video, audio, genetic, meteorological, or geographic — that you want to query without moving it into MaxCompute |
Internal tables
Data in internal tables is stored and managed directly by MaxCompute. Columns support all MaxCompute data types. For the full list, see Data type editions.
External tables
External tables point to data in Object Storage Service (OSS) or Tablestore. MaxCompute stores only the metadata — the underlying data stays where it is. This lets you run SQL queries against unstructured data (video, audio, genetic, meteorological, or geographic files) without ingesting it into MaxCompute.
External tables are available in MaxCompute V2.0 and later.
Manage tables on the DataStudio page of DataWorks, where you can create MaxCompute tables, add tables to favorites, manage data lifecycles, modify table schemas, and configure permissions on tables, resources, and functions.
What's next
-
Table operations — create, delete, and import data into tables
-
External tables overview — create, read from, and write to external tables