Load Methods

The Snowflake destination can load data to Snowflake using the following methods:
COPY command for new data
The COPY command, the default load method, performs a bulk synchronous load to Snowflake, treating all records as INSERTS. Use this method to write new data to Snowflake tables.
The COPY command provides real-time access to data as it is written. It does, however, incur Snowflake warehouse usage fees, which are rounded up to the hour at this time. Use the recommended guidelines to optimize for performance and cost-effectiveness.
Since the COPY command is the default load method, you do not need to configure the Snowflake destination to use this command.
Snowpipe for new data
Snowpipe, the Snowflake continuous ingestion service, performs an asynchronous load to Snowflake, treating all records as INSERTS. Use this method to write new data to Snowflake tables. When needed, you can configure the destination to use a custom Snowflake endpoint.
Snowpipe provides slightly delayed access to data, typically under one minute. Snowpipe incurs Snowflake fees for only the resources used to perform the write.
Before using Snowpipe, perform the prerequisite steps. Also, use the recommended guidelines to optimize for performance and cost-effectiveness.
To use Snowpipe to load new data, enable the Use Snowpipe property on the Snowflake tab of the destination. Then, configure the properties on the Snowpipe tab.
MERGE command for CDC data
Like the COPY command, the MERGE command performs a bulk synchronous load to Snowflake. But instead of treating all records as INSERT, it inserts, updates, upserts, and deletes records as appropriate. Use this method to write change data capture (CDC) data to Snowflake tables using CRUD operations.
Also like the COPY command, the MERGE command provides real-time access to data as it is written. And, it incurs Snowflake warehouse usage fees, which are rounded up to the hour at this time.
Use the recommended guidelines to optimize for performance and cost-effectiveness.
Important: To maintain the original order of data, do not use multiple threads or cluster execution mode when processing CDC data.
To use the MERGE command to load CDC data, select the Processing CDC Data (Use MERGE) property on the Data tab of the destination. Then, enter the Snowflake columns to use as key columns.

For more information about Snowpipe or the COPY or MERGE commands, see the Snowflake documentation.