Snowflake

Source Snowflake #

The following steps refer the extracted Arcion self-hosted CLI download as the $REPLICANT_HOME directory.

Prerequisites for CDC-based replication #

Make sure that you possess the following object privileges for CDC-based replication:

Object Privilege
DATABASE USAGE
SCHEMA USAGE, CREATE
TABLE SELECT, CREATE STREAM, CREATE TABLE

Limitations #

Real-time replication #

  • Streams may become stale over time. For more information, see Data Retention Period and Staleness .
  • Snowflake can extract data on a per-table basis. Therefore, you don’t need to create heartbeat table manually.
Warning: If a stream goes stale, Replicant drops and recreates the stream. This might cause data loss. So we highly recommend that you take necessary measures so that streams don’t become stale.

Native export #

  • Snowflake native export should only be used when the Applier supports file-based bulk loading.
  • We recommend that you use S3 as the stage only when the Applier utilizes S3 as the stage for bulk loading. Otherwise, replication performs similarly to the NATIVE stage type.
  • When using the CSV file format, make sure that the same native-extract-options exist in both the Extractor and Applier configurations.
  • Parquet files might produce an error with TIMESTAMP_TZ or TIMESTAMP_LTZ data.
  • For all general limitations and notes, see Usage notes for the COPY INTO command.

I. Set up connection configuration #

  1. From $REPLICANT_HOME, navigate to the sample connection configuration file:

    vi conf/conn/snowflake.yaml
    
  2. The configuration file has two parts:

    • Parameters related to source Snowflake server connection.
    • Parameters related to stage configuration.

    Note: All communications with Snowflake happens through port 443, the standard port for HTTPS. So all data is encrypted and secure with SSL by default.
    For connecting to Source Snowflake server, you can choose between two methods for an authenticated connection:

    Basic username and password authentication #

    To connect to Snowflake using basic username and password authentication, you have three options:

    You can store your connection credentials in a secrets management service and tell Replicant to retrieve the credentials. For more information, see Secrets management.

    You can also specify your credentials in plain form in the connection configuration file like the sample below:

    type: SNOWFLAKE
    
    host: SNOWFLAKE_HOSTNAME
    port: PORT_NUMBER 
    warehouse: "WAREHOUSE_NAME"
    
    username: "USERNAME"
    password: "PASSWORD"
    
    max-connections: 20 #Specify the maximum number of connections replicant can open in Snowflake
    max-retries: 10
    retry-wait-duration-ms: 1000
    

    Replace the following:

    • SNOWFLAKE_HOSTNAME: The Snowflake hostname. To find your Snowflake hostname, follow these steps:
      1. Go to the Snowflake web interface and sign in into your account.
      2. Click the Account selector toolbar in the bottom of the left navigation menu.
      3. Hover the mouse over your account and click Copy account URL in the items list.
    • PORT_NUMBER: The port number of Snowflake host.
    • WAREHOUSE_NAME: The name of the Snowflake warehouse.
    • USERNAME: The username to connect to the Snowflake server.
    • PASSWORD: The password associated with USERNAME.

    Replicant supports consuming username and password configurations from a credentials store rather than having users specify them in plain text configuration file. You can use KeyStores to store your credentials related to your Snowflake server connections. Use the credential-store parameter for the credentials store details.

    The following parameters are available under credential-store:

    • type: Type of the keystore. Allowed types are PKCS12, JKS, and JCEKS.
    • path : Location of the key-store.
    • key-prefix: You should create entries in the credential store for your configs using a prefix and specify the prefix here. For example, you can create keystore entries with aliases snowflake1_username and snowflake1_password. You can then specify the prefix here as snowflake1_.
    • password: This field is optional. If you don’t specify the keystore password here, then you must use the UUID from your license file as the keystore password. Remember to keep your license file somewhere safe in order to keep the password secure.

    Use RSA key pair for authentication #

    You can also choose to use Snowflake’s key pair authentication support for enhanced authentication security instead of using basic authentication via username and password.

    To set up key pair authentication using RSA keys, follow the steps below:

    Generate the private key

    From your command line, execute the following command to generate an encrypted private key:

    openssl genrsa 2048 | openssl pkcs8 -topk8 -inform PEM -v1 PBE-SHA1-RC4-128 -out rsa_key.p8
    

    The command generates a private key in PEM format:

    -----BEGIN ENCRYPTED PRIVATE KEY-----
    MIIFHDBOBgkqhkiG9w0BBQ0wQTApBgkqhkiG9w0BBQwwHAQIK0h8dqI1n5sCAggA
    MAwGCCqGSIb3DQIJBQAwFAYIKoZIhvcNAwcECNDwqMf6Xx1pBIIEyNmf044S+pEQ
    ...
    -----END ENCRYPTED PRIVATE KEY-----
    

    Important: The command above to generate an encrypted key prompts for a passphrase to grant access to the key. We recommend using a passphrase that complies with PCI DSS standards to protect the generated private key. Additionally, we recommend storing the passphrase in a secure location. When using an encrypted key to connect to Snowflake, you will need to input the passphrase during the initial connection. The use of the passphrase is only for protecting the private key; it’s never to sent to Snowflake servers.

    To generate a long and complex passphrase based on PCI DSS standards, follow the steps below:

    • Go to the PCI Security Standards Document Library.
    • For PCI DSS, select the most recent version and your desired language.
    • Complete the form to access the document.
    • Search for Passwords/passphrases must meet the following: and follow the recommendations for password/passphrase requirements, testing, and guidance.
    Generate a public key

    From the command line, generate the public key by referencing the private key. The following command references the private key contained in a file named rsa_key.p8 created in the previous step:

    openssl rsa -in rsa_key.p8 -pubout -out rsa_key.pub
    

    The command generates a public key in PEM format:

    -----BEGIN PUBLIC KEY-----
    MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAslwTa+Lj5SMI58GiDzWq
    0rwj4FBymfKzHT16RXecnMcx7uI6KsVpqfh9HH0FMb/3C6YEMeGPkaRmKvXYjM5s
    ...
    -----END PUBLIC KEY-----
    
    Store the private and public Keys securely

    After following the above steps, you should find the private and public key files saved in a local directory of your system. Note down the path to those files. The private key is stored using the PKCS#8 (Public Key Cryptography Standards) format and is encrypted using the passphrase you specified in the first step.

    However, maintain caution in protecting the file from unauthorized access using the file permission mechanism provided by your operating system. It’s your responsibility to secure the file when it’s not being used.

    Assign the public key to a Snowflake user

    Execute the following command to assign the public key to a Snowflake user.

    alter user jsmith set rsa_public_key='MIIBIjANBgkqh...';
    
    • Only security administrators (i.e. users with the SECURITYADMIN role) or higher can alter a user.
    • Exclude the public key delimiters in the SQL statement.
    Verify the user's public key fingerprint

    Execute the following command to verify the user’s public key:

    DESC USER jsmith;
    

    The command output is similar to the following:

    +---------------------+-----------------------------------------------------+---------+----------------------------------------------+
    | property            | value                                               | default | description                                  |
    +---------------------+-----------------------------------------------------+---------+----------------------------------------------+
    | NAME                | JSMITH                                              | null    | Name                                         |
    ...
    ...
    | RSA_PUBLIC_KEY      | MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAslwT... | null    | RSA public key of the user                   |
    | RSA_PUBLIC_KEY_FP   | SHA256:nvnONUsfiuycCLMXIEWG4eTp4FjhVUZQUQbNpbSHXiA= | null    | Fingerprint of user's RSA public key.        |
    | RSA_PUBLIC_KEY_2    | null                                                | null    | Second RSA public key of the user            |
    | RSA_PUBLIC_KEY_2_FP | null                                                | null    | Fingerprint of user's second RSA public key. |
    ...
    +---------------------+-----------------------------------------------------+---------+----------------------------------------------+
    
    Edit the connection configuration file

    You need to modify Replicant’s connection configuration file for Snowflake and include RSA key information there. Specifically, add the following parameters to the connection configuration file:

    private-key-path: "/PATH_TO_GENERATED_KEY/rsa_key.p8"
    private-key-passphrase: "PRIVATE_KEY_PASSPHRASE"
    

    Replace the following:

    • PATH_TO_GENERATED_KEY: the local directory path to the rsa_key.p8 keyfile
    • PRIVATE_KEY_PASSPHRASE: the private key passphrase you specified in the first step
    If you specify the private-key-path and private-key-passphrase parameters, you don’t need to specify the password parameter in the connection configuration file.

    For COPY extraction method, you need to specify a stage connection configuration. To specify the stage configuration, use the stage field in the connection configuration file. The following configuration options are available:

    type[v21.06.14.1]

    The stage type. The following stages are supported:

    NATIVE

    Snowflake’s native stage. This stage is created based on table name and job ID. You can specify CSV or PARQUET as the file-format.

    S3

    This specifies S3 as the external stage type, allowing Snowflake to export CSV or Parquet files directly to an S3 bucket.

    To be able to connect to the S3 bucket, you need to provide the connection configuration using the root-dir, conn-url, key-id, and secret-key paramters. Keep in mind that you need to provide the same stage connection configuration in the target connection configuration file. This allows the Applier to pick up these files directly from S3.

    We recommended S3 stage when both the Extractor and the Applier support S3.

    file-format

    The file format to use for the exported data. The following file foramts are supported:

    • CSV
    • PARQUET

    root-dir

    Specifies a directory on stage that can be used to stage bulk-load files.

    conn-url [v21.06.14.1]

    Specifies the URL for the stage. For example, for S3 stage, specify the S3 bucket name.

    key-id

    Specifies the access key ID for AWS account hosting S3.

    secret-key [v21.06.14.1]

    Specifies the secret access Key for AWS account hosting S3. Applies to S3 stage type only.

    Example configuration for NATIVE stage #

    stage:
      type: NATIVE
      file-format: PARQUET
    

    Example configuration for S3 stage #

    stage:
      type: S3
      root-dir: "test_snowflake"      
      conn-url: "replicate-stage"     
      key-id: "AKIAIOSFODNN7EXAMPLE"         
      secret-key: "wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY"      
      file-format: PARQUET            
    

II. Set up Extractor configuration #

To configure replication mode according to your requirements, specify your configuration in the Extractor configuration file. You can find a sample Extractor configuration file snowflake.yaml in the $REPLICANT_HOME/conf/src directory. For example:

Configure snapshot replication #

Snowflake supports two types of extraction method:

  • QUERY: The default JDBC-based extraction method.
  • COPY: The file-based extraction method that uses Snowflake’s native export capability. For more information, see the following section.

Native export [v23.06.30.2] #

From version 23.06.30.2, Arcion Replicant supports exporting Snowflake data into CSV or Parquet files. You can store these files locally or in a remote directory like an S3 bucket. This feature is currently supported in snapshot mode.

To enable native export, follow these steps:

1. Set the extraction method and options in the Extractor configuration file
  • Set the extraction-method parameter to COPY in the Extractor configuration file. This enables Snowflake native export by using the COPY command to export data.

    Extraction method defaults to QUERY and native export is disabled.

  • Set the native-extract-options options in the Extractor configuration file.

    This configuration only applies when you use CSV as the file format for Snowflake native export. This allows you to tune parameters such as the compression type, control characters, delimiter, escape character, and line ending. Make sure to specify similar configurations in the Applier bulk-load parameter to avoid compatibility issues.

You can specify these options both globally and for specific tables.

2. Specify stage configuration in the connection configuration file

Snowflake dumps extracted files into a stage in CSV or Parquet format. To specify the stage configuration, see Parameters related to stage configuration.

Sample snapshot mode configuration #

With native export disabled
snapshot:
  threads: 32
  fetch-size-rows: 100000
  min-job-size-rows: 1000000
  max-jobs-per-chunk: 32
  _traceDBTasks : true

  per-table-config:
    - catalog: DEMO_DB
      schema: tpch
      tables:
        CUSTOMER:
          num-jobs: 32
          split-key: C_CUSTKEY
          row-identifier-key: [ C_CUSTKEY ]
        ORDERS:
          num-jobs: 32
          split-key: O_ORDERKEY
          row-identifier-key: [ O_ORDERKEY ]
        split-hints:
          row-count-estimate: 15000
With native export enabled
snapshot:
  threads: 32
  fetch-size-rows: 100000
  min-job-size-rows: 1000000
  max-jobs-per-chunk: 32
  _traceDBTasks : true

  extraction-method: COPY    

  native-extract-options:
    compression-type: "GZIP"
    control-chars:
      delimiter: ','
      escape: "\\"
      line-end: "\n"
For more information about the configuration parameters for snapshot mode, see Snapshot mode.

Configure realtime replication #

Sample realtime mode configuration #

realtime:
  threads: 32
  fetch-size-rows: 100000
  _traceDBTasks: true
  fetch-interval-s: 0

For more information about the configuration parameters for realtime mode, see Realtime mode.