How to Import Amazon S3 Data to Snowflake

In this article, we’ll show how to load JSON data into Snowflake from Amazon S3.

(This article is part of our Snowflake Guide. Use the right-hand menu to navigate.)

Snowflake and JSON files

Snowflake is a data warehouse on AWS. The Snowflake COPY command lets you copy JSON, XML, CSV, Avro, Parquet, and XML format data files.

But to say that Snowflake supports JSON files is a little misleading—it does not parse these data files, as we showed in an example with Amazon Redshift. Instead, Snowflake copies the entirety of the data into one Snowflake column of type variant. Then you run JSON SQL queries against that.

We will load two data files, which you can download from here:

• Customers
• Orders

Note: This is in NDJSON format. That means the entire file is not a valid JSON file. Instead it is composed of individual JSON records.

Create database and warehouse

We are running our Snowflake cluster on Amazon AWS. (It is not listed as a service on the Amazon AWS Console. Instead you sign up for it on the Snowflake site, then it launches an instance on Amazon, Microsoft, or Google clouds.)

You create a warehouse like this. Here is where you pick the machine site and number of servers, thus picking the computing power and cost.