Snowpark Migration Accelerator : Configuration de la conversion

Lorsque vous lancez pour la première fois l’outil Snowpark Migration Accelerator (SMA), vous devez soit créer un nouveau projet, soit ouvrir un projet existant. Chaque projet peut stocker plusieurs exécutions de SMA pour les phases d’évaluation et de conversion. Après avoir terminé la phase d’évaluation, vous devrez configurer votre projet pour la phase de conversion.

Page de configuration de la conversion

Au cours du processus de conversion, vous disposez de plusieurs options de configuration, bien que la plupart des paramètres par défaut devraient fonctionner correctement dans la plupart des cas.

On the Conversion settings page, choose whether to run the conversion using Default Settings or to select Customize settings to configure advanced options.

Conversion settings page

If you select Customize settings, SMA opens a Conversion settings dialog where you can review and update settings and then click Save settings.

Conversion settings dialog

Conversion Settings

With the following settings from the user interface, you can more finely control how the SMA performs conversion.

  • Pandas

    Convert Pandas API to Snowpark API - Specifies to automatically convert Pandas code to the Snowpark equivalent Pandas API (Snowpark Pandas). When enabled, the tool transforms any Pandas operations it finds in your code into their Snowpark counterparts.

  • DBX

    Convert DBX notebooks to Snowflake notebooks - Specifies to convert the .dbc into Jupyter files in a new folder with the .dbc name.

    Note

    When exporting notebooks, consider exporting them as Databricks, rather than Jupyter. When Jupyter files contain different sources than Python, SMA behavior may be unexpected.

  • Checkpoints

    • Identify and collect checkpoints - Activates the feature.

    • Collect checkpoints as active - Specifies to execute the collected checkpoint in VS Code when running the workload.

    • Collect user-defined functions returning data frame type - Specifies to validate that dataframes should be collected if the user has their own functions that return DataFrames.

    • Mode - Specifies the mode type to validate (Schema or DataFrame).

    • Sample - Specifies the sampling percentage of each DataFrame to validate.

    • Relevant PySpark functions to collect - Specifies the PySpark packages to collect (by default, all of them are checked). You can also add more packages by adding the package’s full name.

Configuration terminée

Once your setup is complete, click the Continue button. This action will initiate the SMA Conversion processes. A progress screen will display the current status of your conversion.

Conversion terminée

After the conversion finishes, SMA automatically displays the Conversion Results screen.