Mapping

tasks process data based on the data flow logic defined in a mapping.

A mapping reads data from one or more sources, transforms the data based on logic that you define, and writes it to one or more targets. Create a mapping when you need to augment or manipulate your data before you load it to a target. For example, if you need to aggregate data, calculate values, perform complex joins, normalize data, or route data to different targets, you can create a mapping to do this.

A

mapping

task runs the data flow logic that you've defined in the mapping. Choose this task type after you've created a mapping so that you can run the data flow logic defined in the mapping.

Data transfer

tasks move data from one or two sources to a target. You can also choose to sort and filter the data before you load it to the target.

Choose this task type when you want to transfer data from a source object, optionally add fields from a second source object, and write the data to a new or existing target object without changing the source data. For example, if you want to move customer records from an on-premises database table to a table in your cloud data warehouse, create a

data transfer

task.

Data loader

tasks provide secure data loading from multi-object sources to corresponding objects in your cloud data warehouse. They can load data incrementally and provide support for schema drift.

To optimize performance, data loading occurs in parallel batches. To fine-tune the data, you can exclude certain objects and fields and also apply some simple filters. If your source data changes frequently, you can load only new and changed records each time the task runs.

Choose this task type when you need to ingest data as-is from multiple objects into your cloud data warehouse. For example, if you need to repeatedly load all the data from files an Amazon S3 bucket to corresponding tables in Snowflake Data Cloud, create a

data loader

task.