Présentation de Snowflake Execution for Desktop

Nouveautés   |   Alex Gnibus   |   28 juin 2023

In a lot of ways, data analytics should be like an all-you-can-eat buffet: You need freedom of choice, and you need scalability (so you can add more plates as you go!). You should be able to choose from a variety of tools while being able to expand compute resources and use cases as needed.

Snowflake and Alteryx are like a data analytics buffet because the Snowflake + Alteryx stack gives you both freedom and scalability. We share the common goal of empowering people with flexible options for working with data. To that end, we’re excited to share a new capability that gives Alteryx customers even more ways to solve analytical problems with data in Snowflake.

Today, we’re announcing Snowflake Execution for Desktop in private preview, giving users the ability to create an analytic workflow in Designer and run the entire workflow in Snowflake.

It’s simple: Build a workflow in Alteryx Designer Desktop’s intuitive interface. Publish the workflow to the cloud. Then schedule and run the workflow in Snowpark Container Services, which enables customers to run containers in Snowflake.

Snowpark Container Services, now available in private preview, is an expansion of Snowflake’s processing engine that provides developers the flexibility to deploy container images in Snowflake managed infrastructure. Data science and engineering teams can run more complex logic inside Snowflake, including a wider range of AI and ML jobs, APIs, and most components of an application, including their web-based interface.

With Alteryx, you don’t have to code to take advantage of the advanced capabilities in Snowpark Container Services. Using Alteryx Designer’s drag-and-drop tools, you can do everything from joining disparate data sets to building a linear regression model — all inside your Snowflake account, so there’s no need to move any data that is already secure and governed.

How it works

Alteryx recently expanded its cloud-connected experiences, which includes Cloud Execution for Desktop. This enables users to create a workflow on Alteryx Designer, save their work to the Alteryx Analytics Cloud Platform, and run their workflow in the cloud.

Snowflake Execution for Desktop uses this capability to bring the Alteryx engine to the Snowflake ecosystem. When you run a Designer workflow in Snowpark Container Services, Alteryx initiates the job and executes it in your Snowflake account. The Alteryx AMP engine runs the workflow in Snowflake, keeping the data in Snowflake.
What does this process look like? It’s as easy as three steps:

  1. Build the workflow in Designer (Desktop) using drag-and-drop tools, custom code, or a combination of both — it’s all up to the user.
  2. Save the workflow as a package and publish to the Alteryx Analytics Cloud Platform.
  3. Schedule and run the workflow in Snowflake.

When you execute a Designer workflow in Snowpark Container Services, you can use any of the 160+ tools supported on the Linux build of the AMP engine, including predictive analytics tools like Linear Regression, Neural Network, and Time Series Forecast. Like a buffet: Choose from a massive menu of no-code and code-friendly building blocks and power through your data with Snowflake’s scalable, powerful containers.

What it solves

Snowflake Execution for Desktop makes it easy and secure for employees to get their hands on the data they need, transform that data, and unlock advanced use cases.
Empower both no-code and code-friendly approaches: Alteryx empowers employees across diverse backgrounds and skill sets to utilize Snowflake for data transformation and advanced analytics use cases. Because Alteryx enables both no-code and code-friendly tools, Snowflake Execution for Desktop makes it possible for employees to collaborate on a workflow and fully run it in Snowflake.

  1. Get richer data insights at a faster pace: Combine data sets easily in Designer and run the entire workflow in Snowflake, driving efficiency for AI/ML models and data science use cases.
  2. Leverage a wide range of transformation capabilities directly in Snowflake: Many types of transformations require moving data out of Snowflake, transforming the data in Alteryx Designer, then returning the data to Snowflake. Now, you can perform 160+ transformations in Alteryx Designer without moving the data outside the Snowflake security perimeter.
  3. This new capability is a great complement to Designer’s generally available in-database pushdown processing capabilities with Snowflake. Pushdown processing is especially helpful for processing large volumes of data simultaneously and performing SQL-based transformations like filtering and joining data sets.

Use case example: Predicting buffet demand

Remember when we said Snowflake and Alteryx are like an all-you-can eat buffet for data analytics? Well, here’s a use case involving an actual buffet!
Have you ever walked into a buffet at a Las Vegas casino and wondered how much food is left over at the end of the night? How do they know how much food people will eat and if the casino can still make a profit on the price of the meal? How many servers should be staffed for any given shift?

If you’re in charge of a casino buffet, you’ll want to predict these things so you can plan for demand. In this example, your casino uses Snowflake to centralize and process data, and you want to create your workflow in Designer with advanced tools, while keeping and transforming the data in Snowflake.

This sounds like a job for Snowpark Container Services and Snowflake Execution for Desktop.

To predict buffet demand, you can use historical attendance data and combine it with holidays, days of the week, and weather data from the Snowflake Data Marketplace. In Designer, you can test these datasets to validate if they influence the seasonal volume of customers and build a model for the prediction. You can build your workflow using the wide range of familiar Designer tools, from preparing data and performing data investigation to building a predictive modeland outputting the scored data set in a PDF summary — and run it all in Snowflake.

Check out this video to see a demo of this use case and learn more about Snowflake Execution for Desktop.

Learn more about Snowpark Container Services. And start solving today with the Analytics Cloud Platform free trial.

Tags