MaxCompute MapReduce jobs read from and write to MaxCompute tables.
Supported data types
MapReduce supports the following MaxCompute built-in data types for both input and output:
| Category | Types |
|---|---|
| Numeric | BIGINT, DOUBLE |
| String | STRING |
| Date and time | DATETIME |
| Boolean | BOOLEAN |
User-defined data types (UDTs) are not supported.
Input
| Capability | Details |
|---|---|
| Multiple table input | A job can read from multiple tables with different schemas. Use the map function to identify which table the current record belongs to. |
| Null values | Accepted as input. |
| Views | Not supported as input. |
Output
A reduce job can write to:
-
Multiple different tables
-
Multiple partitions within the same table
Target tables can have different schemas. Use labels to distinguish between multiple outputs. The default output has no label.
MapReduce does not support functions that produce no output.
Example: multiple inputs and outputs
For a working example that demonstrates multi-table input and label-based output routing, see MultipleInOut.