Deprecated — Snowflake Connector 1.x for Informatica Cloud

This topic contains information about how to set up and use version 1.x of the Snowflake Connector. It explains how Informatica Cloud organization administrators and business users can use the Snowflake Connector to publish data to Snowflake.

The connector implements the Informatica Cloud Connector SDK. It can be deployed on both Informatica Cloud and Informatica PowerCenter 9.6.1. For assistance deploying the connector on PowerCenter, please contact Snowflake Support.

Note

Snowflake also provides an ODBC library that can be used for data integration with Informatica’s products; currently, this library only supports read functionality.

In this Topic:

Introduction to Snowflake Connector

Snowflake Connector Overview

Snowflake provides programmatic APIs for querying and modifying data in the form of industry-standard ODBC and JDBC libraries. The ODBC library can be used with Informatica products using standard ODBC connectors. See the Informatica documentation for configuring the ODBC connector. The ODBC library can be downloaded from your Snowflake account. However, writing or updating large volumes of data into Snowflake using ODBC is often not the most efficient or effective way to perform these operations.

The Snowflake Connector is designed to improve throughput of bulk insertion, modification, and deletion of large numbers of rows in Snowflake. It works by caching row-by-row data it receives through Informatica, uploading it asynchronously to cloud storage in the form of compressed character-delimited files, and importing data from the files using the Snowflake COPY command.

Snowflake Connector Implementation

Data submitted for processing is staged within the internal stage for the configured connection user (identified by the ~ character).

Subdirectories are created in the stage for each job. Multiple batches can be processed, with a corresponding subdirectory within the user stage for each batch. Each subdirectory includes the following information:

  • Name of the target table.

  • Name of the operation (INSERT, DELETE, UPSERT, MODIFY).

  • Timestamp and a unique identifier (consecutive number).

In the History History tab page in the Snowflake web interface, the following commands are displayed for the user configured to run the process:

  • SQL statements configured to be run before a job.

  • Sequence of PUT commands to upload data files to the stage.

  • Creation of a temporary table to stage data.

  • COPY command to import data into the staging table, optionally in validation mode to first identify/retrieve data conversion errors.

  • DELETE, MERGE or INSERT command to process the data.

  • RM command to clean up staged files from the stage.

This sequence may be modified by the connector to optimize performance.

Data errors are reported to Informatica to be written to the error file session log, and may terminate the job if it is so configured. On its own, the Snowflake loading process skips all data conversion errors.

Snowflake Connections

Snowflake Connection Overview

The Snowflake Connector uses the Snowflake JDBC driver to connect. The driver library is included in the connector distribution.

Snowflake Connection Properties

The Snowflake Connector uses the following properties for connecting to Snowflake:

USER and PASSWORD

User name and password for the account that will be used for the loading process. Snowflake recommends using a dedicated user with the appropriate write privileges for the table where data will be loaded.

Snowflake URL

JDBC URL for connecting to the Snowflake database and schema in your account. For example:

jdbc:snowflake://xy12345.snowflakecomputing.com/?db=load&schema=etl

Where:

  • xy12345 is the name of your account (provided by Snowflake).

  • If your account is located in a region other than US West, the JDBC connection string must also include the region ID after your account name in the form of <account_id>.<region_id>.snowflakecomputing.com.

  • load is the name of the default database to use for loading data.

  • etl is the name of the schema (in the load database) containing the tables to be loaded.

Note

During design time, metadata browsing is limited to the Snowflake schema and database specified in the connection or in the search path of the user.

Tip

If you run a job that includes a large set of data and very complex transformations, it may take a long time to complete. If the job takes over 4 hours, the Snowflake connection token may expire. To avoid this situation, you can specify the client_session_keep_alive parameter in the JDBC connection string, which prevents the connection token from expiring. For example:

jdbc:snowflake://xy12345.snowflakecomputing.com/?...&client_session_keep_alive=true

Start transaction for jobs

If set, the connector will initiate a transaction before the start of every job, and commit or rollback upon the completion or failure of the job.

Note

Informatica does not support operation rollback or disconnect in the connector API. Terminating a job may leave hanging table locks and an uncommitted transaction that may be needed to be released manually from the Snowflake command line.

Abort on data errors

When this property is selected, every job will stop processing if any data conversion errors are encountered during data import. To rollback partial changes when errors are encountered, also set Start transaction for jobs.

Note

Because data is loaded asynchronously, some data may already be committed if this property is used and more than one batch of data was generated.

Propagate data stream

The connector implements midstream write interface that allows chaining of data processing. If this property is selected, the connector will pass data for further processing.

For better performance, do not select this property.

Snowflake Data Synchronization Tasks

The connector provides advanced target properties for specifying Snowflake-specific actions and properties to use when a data synchronization task is performed.

Snowflake Advanced Target Properties

The following table describes the advanced target properties that can be specified for a data synchronization task:

Advanced Target Property

Description

Update key columns

Semicolon separated list of column names in the target table that should be used as a composite key for DELETE or MODIFY operations.

Execute before

SQL statement that will be executed prior to start of a job.

Truncate table

Delete all data from the target table prior to execution of the job. This statement is completed after execution of the Execute before statement.

Execute after

SQL statement that will be executed after completion of a job.

Process data in one batch

When this property is checked, the connector will upload all data from the job prior to processing it.

Preserve stage file on Error

Preserve staged data file when an error occurs in loading data. This property is valid only if Abort on data error is enabled.

Use Local Timezone

Use agent local timezone to convert TIMESTAMP/datetime data. By default, UTC is used in conversions.

Success File Directory

Not currently used.

Error File Directory

Not currently used.

Database Override

Name of the database to update; overrides the target database defined for the data synchronization task. Do not specify values for database override, schema override, or table override in a Data Synchronization task. You can specify the values in a PowerCenter session.

Schema Override

Name of the schema to update; overrides the target schema defined for the data synchronization task. Do not specify values for database override, schema override, or table override in a Data Synchronization task. You can specify the values in a PowerCenter session.

Table Override

Name of the table to update; overrides the target table defined for the data synchronization task. Do not specify values for database override, schema override, or table override in a Data Synchronization task. You can specify the values in a PowerCenter session.

Usage Notes:

  • Snowflake does not enforce primary or foreign key constraints, and does not preserve metadata for keys. You must specify the Update key columns property even if corresponding columns are marked as key in the Informatica environment.

  • The Process data in one batch property may delay completion of the job, but guarantees that no data will be persisted in case of a failure, and without the overall transaction. Processing maximum amount of data at a time also maximizes utilization of Snowflake warehouse parallelism.

  • The Database Override, Schema Override, and Table Override attributes are used by PowerCenter to provide values at runtime that override the target database, schema, and/or table for the data synchronization task. This enables using the same data synchronization task to update tables in multiple databases and schemas. The fields are blank by default and should be left blank because the values for the attributes are provided at runtime.