Appearance
Data Type Mapping for Parquet Unloads
The following table shows which parquet types are used when data from Yellowbrick tables is unloaded in parquet format. For more details about each type, go to the linked topic.
Note: You cannot modify the mapping of Yellowbrick types to parquet types when data is unloaded with ybunload. Only the following default mappings are supported. For example, even if the range of values in a BIGINT table column would fit a smaller target parquet data type, the target type will always be an INT64 type, as shown in the table.
| Yellowbrick Data Type | Mapped Data Type in Unloaded Parquet Files |
|---|---|
| BOOLEAN | Boolean |
| SMALLINT | INT32 (16 bits, signed) |
| INTEGER | INT32 (32 bits, signed) |
| BIGINT | INT64 (64 bits, signed) |
| DECIMAL | FIXED_LEN_BYTE_ARRAY (ranging from 4 to 16 bytes) |
| REAL | Float |
| DOUBLE PRECISION | Double |
| UUID | FIXED_LEN_BYTE_ARRAY (16 bytes) |
| VARCHAR | String (Binary) |
| CHAR | |
| DATE | INT32 (Date) |
| TIME | INT64 (Time), adjusted to UTC, up to microseconds precision |
| TIMESTAMP | INT64 (Timestamp), adjusted to UTC, up to microseconds precision |
| TIMESTAMP WITH TIME ZONE | |
| IPV4 | String (Binary) |
| IPV6 | |
| MACADDR | |
| MACADDR8 |
Parent topic:Unloading Data to Parquet Files