SnowPipe: Cloud Data Ingestion Tool Powered by Snowflake

4 min readFeb 14, 2024

How does automated real-time data loading redefine cloud engineering efficiency? Enter SnowPipe, Snowflake’s innovative feature, automating data loading upon file appearance in designated buckets, perfect for real-time analytics. Uncover the transformative potential that SnowPipe holds for optimizing your analytics workflow.

By Lakshmi Haswitha Madupalli

Automating Cloud Data Ingestion with Snowpipe

Today, let us will discuss Snowpipe, a very important feature of the cloud data engineering and warehousing platform, Snowflake. This feature allows you to automatically load data as soon as files appear in a particular bucket. When new files are added to the bucket, they are automatically loaded into the specified table. This is especially important when it comes to transactional or event data that you want to be readily available for analysis.

Using SnowPipe to automatically fetch new data as soon as it appears in a particular bucket is a typical use case, also known as a ‘serverless function’. This means that instead of utilizing a virtual warehouse, computing resources are managed by Snowflake itself. Hence, there’s no need to worry about virtual warehouses in this scenario.

Suppose you have a specific bucket, and you can manually copy data from that bucket to your table. However, SnowPipe can automatically detect when a file is loaded or placed into that bucket. Just get that notification and you can set up an event notification to trigger SnowPipe.

S3 Bucket to Snowpipe Services Flowchart | Factspan

Setting up Snowpipe

1) Create Stage

First step is to create a stage object. This includes the connection and location to the external stage to copy the data. After that, we need to create a copy command. We recommend that you also test this copy command after creation.

You can then create a pipe containing this copy command definition. Before using the COPY command in a Snowpipe definition, you should ensure that the COPY command runs successfully and correctly. A stage can be an S3 bucket or an Azure Blob where streaming data arrives continuously.

// Create table first
CREATE OR REPLACE TABLE SNOW_DB.PUBLIC.employees (id INT,first_name STRING,last_name STRING,email STRING,location STRING,department STRING);// Create file format objectCREATE OR REPLACE file format DATABASE.file_formats.csv_fileformattype = csvfield_delimiter = ','skip_header = 1null_if = ('NULL','null')empty_field_as_null = TRUE;// Create stage object with integration object & file format objectCREATE OR REPLACE stage DATABASE.external_stages.csv_folderURL = 's3://snowflakebucket/csv/snowpipe'STORAGE_INTEGRATION = s3_intFILE_FORMAT = DATABASE.file_formats.csv_fileformat;// Create stage object with integration object & file format objectLIST @DATABASE.external_stages.csv_folder;// Create schema to keep things organizedCREATE OR REPLACE SCHEMA DATABASE.pipes;

2) Create Pipe

We need to create a pipe and name the pipe as Employee. After creating a pipeline, it’s still not operational. We need to configure notifications and receive files to initiate the copy process.

// Define pipe
CREATE OR REPLACE pipe DATABASE.pipes.employee_pipeauto_ingest = TRUEASCOPY INTO SNOW_DB.PUBLIC.employeesFROM @DATABASE.external_stages.csv_folder ;// Describe pipeDESC pipe employee_pipe;SELECT * FROM SNOW_DB.PUBLIC.employees ;

AWS Data Engineering Essentials Guidebook

AWS empowers efficient data engineering with S3, Glue, EMR, Kinesis, and Lambda for seamless batch and real-time processing.

3) Configure Notifications

There are a couple of ways to trigger a pipe to continuously load data.

  1. Configure events via cloud platform: This will trigger the pipe when new files arrive. When the event occurs, the associated pipe is also fired to load the data into the target table. This is the preferred option as it is less hassle and more robust than the alternative method.
  2. Trigger Snowpipe using the REST API: Write a custom program that you can run whenever you want Snowpipe to load.

4) Manage Pipes

After setting up the pipeline, it’s crucial to handle operations like pause, resume, refresh, list files in steps, check for a paused channel with PendingFileCount 0, and recreate channels to modify the COPY command in the definition. Additionally, ensure checks for the running status of channels.

// Preparation table first
CREATE OR REPLACE TABLE SNOW_DB.PUBLIC.employees2 (id INT,first_name STRING,last_name STRING,email STRING,location STRING,department STRING);// Pause pipeALTER PIPE DATABASE.pipes.employee_pipe SET PIPE_EXECUTION_PAUSED = true;// Verify pipe is paused and has pendingFileCount 0 SELECT SYSTEM$PIPE_STATUS('DATABASE.pipes.employee_pipe') ;// Recreate the pipe to change the COPY statement in the definitionCREATE OR REPLACE pipe DATABASE.pipes.employee_pipeauto_ingest = TRUEASCOPY INTO SNOW_DB.PUBLIC.employees2FROM @DATABASE.external_stages.csv_folder ;ALTER PIPE DATABASE.pipes.employee_pipe refresh;// List files in stageLIST @DATABASE.external_stages.csv_folder ;SELECT * FROM SNOW_DB.PUBLIC.employees2;// Reload files manually that where aleady in the bucketCOPY INTO SNOW_DB.PUBLIC.employees2FROM @DATABASE.external_stages.csv_folder; // Resume pipeALTER PIPE DATABASE.pipes.employee_pipe SET PIPE_EXECUTION_PAUSED = false;// Verify pipe is running againSELECT SYSTEM$PIPE_STATUS('DATABASE.pipes.employee_pipe') ;

Real-Time Data Acceleration and Advanced ML Integrations

Snowpipe accelerates data loading by processing files in micro-batches, providing real-time access within minutes. Snowflake’s cloud-native design ensures scalability and concurrent access with features like multi-cluster and shared data architecture. Automatic optimization streamlines data warehouse management and ensures efficient query performance.

Upcoming features of cloud data engineering Snowflake is looking to strengthen security measures, enhance data sharing capabilities, and foster collaborative analytics. Snowflake is set to introduce advanced machine learning integrations, improved support for semi-structured data, and expanded real-time analytics capabilities to adapt to our evolving data needs.

Get in touch with Factspan for Cloud Data Engineering Solutions.

This Article is originally published on Factspan




Factspan is a pure play analytics company. We partner with you to build an analytics center of excellence, uncovering insights and solutions from your data.