Snowpark is a unified programming model for building data-intensive applications in Snowflake. It provides a familiar programming experience in Python, Java, and Scala, and it can be deployed on-premises or in the cloud.
Here are the deployment options available for Snowpark:
- Local development: You can install the Snowpark libraries on your local machine and develop applications using your favorite IDE.
- Snowflake Dataproc: You can use Snowflake Dataproc to create a managed Hadoop and Spark cluster in Snowflake. Snowpark is pre-installed on Snowflake Dataproc clusters, so you can start developing applications right away.
- Snowflake Compute Instances: You can create a dedicated Snowflake Compute Instance to run Snowpark applications. This option gives you the most control over your environment, but it also requires more administrative overhead.
- Snowflake Container Services: You can deploy Snowpark applications in containers using Snowflake Container Services. This option gives you the flexibility to run Snowpark applications on any infrastructure, including on-premises or in the cloud.
Yes, Snowpark can be deployed on-premises as well as in the cloud. The deployment options listed above are available for both on-premises and cloud deployments.
Here are some additional considerations for deploying Snowpark on-premises:
- You will need to install the Snowpark libraries on your on-premises servers.
- You will need to configure your on-premises environment to connect to Snowflake.
- You may need to make some changes to your on-premises security policies to allow Snowpark applications to access Snowflake.
Here are some additional considerations for deploying Snowpark in the cloud:
- You will need to create a Snowflake account and create a database.
- You will need to create a Snowflake Compute Instance or use Snowflake Dataproc to create a managed Hadoop and Spark cluster.
- You will need to install the Snowpark libraries on your cloud machine.
- You will need to configure your cloud machine to connect to Snowflake.