Cloud Storage transfers

The BigQuery Data Transfer Service for Cloud Storage allows you to schedule recurring data loads from Cloud Storage to BigQuery.

Before you begin

Before you create a Cloud Storage transfer, do the following:

Limitations

Recurring transfers from Cloud Storage to BigQuery are subject to the following limitations:

  • All files matching the patterns defined by either a wildcard or by runtime parameters for your transfer must share the same schema you defined for the destination table, or the transfer will fail. Table schema changes between runs also causes the transfer to fail.
  • Because Cloud Storage objects can be versioned, it's important to note that archived Cloud Storage objects are not supported for BigQuery transfers. Objects must be live to be transferred.
  • Unlike individual loads of data from Cloud Storage to BigQuery, for ongoing transfers you must create the destination table before setting up the transfer. For CSV and JSON files, you must also define the table schema in advance. BigQuery cannot create the table as part of the recurring data transfer process.
  • Transfers from Cloud Storage set the Write preference parameter to APPEND by default. In this mode, an unmodified file can only be loaded into BigQuery once. If the file's last modification time property is updated, then the file will be reloaded.
  • BigQuery Data Transfer Service does not guarantee all files will be transferred or transferred only once if Cloud Storage files are touched while in mid-transfer. You are subject to the following limitations when you load data into BigQuery from a Cloud Storage bucket:

  • If your dataset's location is set to a value other than the US multi-region, then the Cloud Storage bucket must be in the same region or contained in the same multi-region as the dataset.

  • BigQuery does not guarantee data consistency for external data sources. Changes to the underlying data while a query is running can result in unexpected behavior.

  • BigQuery does not support Cloud Storage object versioning. If you include a generation number in the Cloud Storage URI, then the load job fails.

  • Depending on the format of your Cloud Storage source data, there might be additional limitations. For more information, see:

  • Your Cloud Storage bucket must be in a location that is compatible with the region or multi-region of the destination dataset in BigQuery. This is known as colocation. See Cloud Storage transfer Data locations for details.

Minimum intervals

  • Source files are picked up for transfer immediately, with no minimum file age.
  • The minimum interval time between recurring transfers is 15 minutes. The default interval for a recurring transfer is every 24 hours.

Required permissions

When you load data into BigQuery, you need permissions that allow you to load data into new or existing BigQuery tables and partitions. If you are loading data from Cloud Storage, you'll also need access to the bucket that contains your data. Ensure that you have the following required permissions:

  • BigQuery: Ensure that the person or service account creating the transfer has the following permissions in BigQuery:

    • bigquery.transfers.update permissions to create the transfer
    • Both bigquery.datasets.get and bigquery.datasets.update permissions on the target dataset

    The bigquery.admin predefined IAM role includes bigquery.transfers.update, bigquery.datasets.update and bigquery.datasets.get permissions. For more information on IAM roles in the BigQuery Data Transfer Service, see Access control.

  • Cloud Storage: storage.objects.get permissions are required on the individual bucket or higher. If you are using a URI wildcard, you must also have storage.objects.list permissions. If you would like to delete the source files after each successful transfer, you also need storage.objects.delete permissions. The storage.objectAdmin predefined IAM role includes all of these permissions.

Set up a Cloud Storage transfer

To create a Cloud Storage transfer in the BigQuery Data Transfer Service:

Console

  1. Go to the BigQuery page in the Google Cloud console.

    Go to the BigQuery page

  2. Click Data transfers.

  3. Click Create transfer.

  4. In the Source type section, for Source, choose Google Cloud Storage.

    Transfer source

  5. In the Transfer config name section, for Display name, enter a name for the transfer such as My Transfer. The transfer name can be any value that lets you identify the transfer if you need to modify it later.

    Transfer name

  6. In the Schedule options section:

    • Select a Repeat frequency. If you select Hours, Days, Weeks, or Months, you must also specify a frequency. You can also select Custom to specify a custom repeat frequency. If you select On-demand, then this transfer runs when you manually trigger the transfer.

    • If applicable, select either Start now or Start at set time and provide a start date and run time.

  7. In the Destination settings section, for Destination dataset, choose the dataset you created to store your data.

    Transfer dataset

  8. In the Data source details section:

    1. For Destination table, enter the name of your destination table. The destination table must follow the table naming rules. Destination table names also support parameters.
    2. For Cloud Storage URI, enter the Cloud Storage URI. Wildcards and parameters are supported.
    3. For Write preference, choose:
      • APPEND to incrementally append new data to your existing destination table. APPEND is the default value for Write preference.
      • MIRROR to overwrite data in the destination table during each transfer run.

    For more information about how BigQuery Data Transfer Service ingests data using either APPEND or MIRROR, see Data ingestion for Cloud Storage transfers. For more information about the writeDisposition field, see JobConfigurationLoad.

    1. For Delete source files after transfer, check the box if you want to delete the source files after each successful transfer. Delete jobs are best effort. Delete jobs do not retry if the first effort to delete the source files fails.
    2. In the Transfer Options section:

      1. Under All Formats:
        1. For Number of errors allowed, enter the maximum number of bad records that BigQuery can ignore when running the job. If the number of bad records exceeds this value, an 'invalid' error is returned in the job result, and the job fails. The default value is 0.
        2. (Optional) For Decimal target types, enter a comma-separated list of possible SQL data types that the source decimal values could be converted to. Which SQL data type is selected for conversion depends on the following conditions:
          • The data type selected for conversion will be the first data type in the following list that supports the precision and scale of the source data, in this order: NUMERIC, BIGNUMERIC, and STRING.
          • If none of the listed data types will support the precision and the scale, the data type supporting the widest range in the specified list is selected. If a value exceeds the supported range when reading the source data, an error will be thrown.
          • The data type STRING supports all precision and scale values.
          • If this field is left empty, the data type will default to "NUMERIC,STRING" for ORC, and "NUMERIC" for the other file formats.
          • This field cannot contain duplicate data types.
          • The order of the data types that you list in this field is ignored.
      2. Under JSON, CSV:
        • For Ignore unknown values, check the box if you want the transfer to drop data that does not fit the destination table's schema.
      3. Under AVRO:
        • For Use avro logical types, check the box if you want the transfer to convert Avro logical types to their corresponding BigQuery data types. The default behavior is to ignore the logicalType attribute for most of the types and use the underlying Avro type instead.
      4. Under CSV:

        1. For Field delimiter, enter the character that separates fields. The default value is a comma.
        2. For Quote character, enter the character that is used to quote data sections in a CSV file. The default value is a double-quote (").
        3. For Header rows to skip, enter the number of header rows in the source file(s) if you don't want to import them. The default value is 0.
        4. For Allow quoted newlines, check the box if you want to allow newlines within quoted fields.
        5. For Allow jagged rows, check the box if you want to allow the transfer of rows with missing NULLABLE columns.
  9. In the Service Account menu, select a service account from the service accounts associated with your Google Cloud project. You can associate a service account with your transfer instead of using your user credentials. For more information about using service accounts with data transfers, see Use service accounts.

    • If you signed in with a federated identity, then a service account is required to create a transfer. If you signed in with a Google Account, then a service account for the transfer is optional.
    • The service account must have the required permissions for both BigQuery and Cloud Storage.
  10. Optional: In the Notification options section:

    1. Click the toggle to enable email notifications. When you enable this option, the owner of the transfer configuration receives an email notification when a transfer run fails.
    2. For Select a Pub/Sub topic, choose your topic name or click Create a topic. This option configures Pub/Sub run notifications for your transfer.
  11. Optional: In the Advanced options section:

    • If you use CMEKs, select Customer-managed key. A list of your available CMEKs appears for you to choose from.

    For information about how CMEKs work with the BigQuery Data Transfer Service, see Specify encryption key with transfers.

  12. Click Save.

bq

Enter the bq mk command and supply the transfer creation flag — --transfer_config. The following flags are also required:

  • --data_source
  • --display_name
  • --target_dataset
  • --params

Optional flags:

  • --destination_kms_key: Specifies the key resource ID for the Cloud KMS key if you use a customer-managed encryption key (CMEK) for this transfer. For information about how CMEKs work with the BigQuery Data Transfer Service, see Specify encryption key with transfers.
  • --service_account_name: Specifies a service account to use for Cloud Storage transfer authentication instead of your user account.
bq mk \
--transfer_config \
--project_id=PROJECT_ID \
--data_source=DATA_SOURCE \
--display_name=NAME \
--target_dataset=DATASET \
--destination_kms_key="DESTINATION_KEY" \
--params='PARAMETERS' \
--service_account_name=SERVICE_ACCOUNT_NAME

Where:

  • PROJECT_ID is your project ID. If --project_id isn't supplied to specify a particular project, the default project is used.
  • DATA_SOURCE is the data source, for example, google_cloud_storage.
  • NAME is the display name for the transfer configuration. The transfer name can be any value that lets you identify the transfer if you need to modify it later.
  • DATASET is the target dataset for the transfer configuration.
  • DESTINATION_KEY: the Cloud KMS key resource ID—for example, projects/project_name/locations/us/keyRings/key_ring_name/cryptoKeys/key_name.
  • PARAMETERS contains the parameters for the created transfer configuration in JSON format. For example: --params='{"param":"param_value"}'.
    • destination_table_name_template: the name of the destination BigQuery table.
    • data_path_template: the Cloud Storage URI that contains your files to be transferred, which can include one wildcard.
    • write_disposition: determines if matching files are appended to the destination table or mirrored entirely. The supported values are APPEND or MIRROR. For information about how the BigQuery Data Transfer Service appends or mirrors data in Cloud Storage transfers, see Data ingestion for Cloud Storage transfers.
    • file_format: the format of the files that you want to transfer. The format can be CSV, JSON, AVRO, PARQUET, or ORC. The default value is CSV.
    • max_bad_records: for any file_format value, the maximum number of bad records that can be ignored. The default value is 0.
    • decimal_target_types: for any file_format value, a comma-separated list of possible SQL data types that the source decimal values could be converted to. If this field is not provided, the data type defaults to "NUMERIC,STRING" for ORC, and "NUMERIC" for the other file formats.
    • ignore_unknown_values: for any file_format value, set to TRUE to accept rows that contain values that don't match the schema. For more information, see the ignoreUnknownvalues field details in the JobConfigurationLoad reference table.
    • use_avro_logical_types: for AVRO file_format values, set to TRUE to interpret logical types into their corresponding types (for example, TIMESTAMP), instead of only using their raw types (for example, INTEGER).
    • parquet_enum_as_string: for PARQUET file_format values, set to TRUE to infer PARQUET ENUM logical type as STRING instead of the default BYTES.
    • parquet_enable_list_inference: for PARQUET file_format values, set to TRUE to use schema inference specifically for PARQUET LIST logical type.
    • reference_file_schema_uri: a URI path to a reference file with the reader schema.
    • field_delimiter: for CSV file_format values, a character that separates fields. The default value is a comma.
    • quote: for CSV file_format values, a character that is used to quote data sections in a CSV file. The default value is a double-quote (").
    • skip_leading_rows: for CSV file_format values, indicate the number of leading header rows that you don't want to import. The default value is 0.
    • allow_quoted_newlines: for CSV file_format values, set to TRUE to allow newlines within quoted fields.
    • allow_jagged_rows : for CSV file_format values, set to TRUE to accept rows that are missing trailing optional columns. The missing values are filled in with NULL.
    • preserve_ascii_control_characters: for CSV file_format values, set to TRUE to preserve any embedded ASCII control characters.
    • encoding: specify the CSV encoding type. Supported values are UTF8, ISO_8859_1, UTF16BE, UTF16LE, UTF32BE, and UTF32LE.
    • delete_source_files: set to TRUE to delete the source files after each successful transfer. Delete jobs don't rerun if the first try to delete the source file fails. The default value is FALSE.
  • SERVICE_ACCOUNT_NAME is the service account name used to authenticate your transfer. The service account should be owned by the same project_id used to create the transfer and it should have all of the required permissions.

For example, the following command creates a Cloud Storage transfer named My Transfer using a data_path_template value of gs://mybucket/myfile/*.csv, target dataset mydataset, and file_format CSV. This example includes non-default values for the optional params associated with the CSV file_format.

The transfer is created in the default project:

bq mk --transfer_config \
--target_dataset=mydataset \
--display_name='My Transfer' \
--params='{"data_path_template":"gs://mybucket/myfile/*.csv",
"destination_table_name_template":"MyTable",
"file_format":"CSV",
"max_bad_records":"1",
"ignore_unknown_values":"true",
"field_delimiter":"|",
"quote":";",
"skip_leading_rows":"1",
"allow_quoted_newlines":"true",
"allow_jagged_rows":"false",
"delete_source_files":"true"}' \
--data_source=google_cloud_storage

After running the command, you receive a message like the following:

[URL omitted] Please copy and paste the above URL into your web browser and follow the instructions to retrieve an authentication code.

Follow the instructions and paste the authentication code on the command line.

API

Use the projects.locations.transferConfigs.create method and supply an instance of the TransferConfig resource.

Java

Before trying this sample, follow the Java setup instructions in the BigQuery quickstart using client libraries. For more information, see the BigQuery Java API reference documentation.

To authenticate to BigQuery, set up Application Default Credentials. For more information, see Set up authentication for client libraries.

import com.google.api.gax.rpc.ApiException;
import com.google.cloud.bigquery.datatransfer.v1.CreateTransferConfigRequest;
import com.google.cloud.bigquery.datatransfer.v1.DataTransferServiceClient;
import com.google.cloud.bigquery.datatransfer.v1.ProjectName;
import com.google.cloud.bigquery.datatransfer.v1.TransferConfig;
import com.google.protobuf.Struct;
import com.google.protobuf.Value;
import java.io.IOException;
import java.util.HashMap;
import java.util.Map;

// Sample to create google cloud storage transfer config
public class CreateCloudStorageTransfer {

  public static void main(String[] args) throws IOException {
    // TODO(developer): Replace these variables before running the sample.
    final String projectId = "MY_PROJECT_ID";
    String datasetId = "MY_DATASET_ID";
    String tableId = "MY_TABLE_ID";
    // GCS Uri
    String sourceUri = "gs://cloud-samples-data/bigquery/us-states/us-states.csv";
    String fileFormat = "CSV";
    String fieldDelimiter = ",";
    String skipLeadingRows = "1";
    Map<String, Value> params = new HashMap<>();
    params.put(
        "destination_table_name_template", Value.newBuilder().setStringValue(tableId).build());
    params.put("data_path_template", Value.newBuilder().setStringValue(sourceUri).build());
    params.put("write_disposition", Value.newBuilder().setStringValue("APPEND").build());
    params.put("file_format", Value.newBuilder().setStringValue(fileFormat).build());
    params.put("field_delimiter", Value.newBuilder().setStringValue(fieldDelimiter).build());
    params.put("skip_leading_rows", Value.newBuilder().setStringValue(skipLeadingRows).build());
    TransferConfig transferConfig =
        TransferConfig.newBuilder()
            .setDestinationDatasetId(datasetId)
            .setDisplayName("Your Google Cloud Storage Config Name")
            .setDataSourceId("google_cloud_storage")
            .setParams(Struct.newBuilder().putAllFields(params).build())
            .setSchedule("every 24 hours")
            .build();
    createCloudStorageTransfer(projectId, transferConfig);
  }

  public static void createCloudStorageTransfer(String projectId, TransferConfig transferConfig)
      throws IOException {
    try (DataTransferServiceClient client = DataTransferServiceClient.create()) {
      ProjectName parent = ProjectName.of(projectId);
      CreateTransferConfigRequest request =
          CreateTransferConfigRequest.newBuilder()
              .setParent(parent.toString())
              .setTransferConfig(transferConfig)
              .build();
      TransferConfig config = client.createTransferConfig(request);
      System.out.println("Cloud storage transfer created successfully :" + config.getName());
    } catch (ApiException ex) {
      System.out.print("Cloud storage transfer was not created." + ex.toString());
    }
  }
}

Specify encryption key with transfers

You can specify customer-managed encryption keys (CMEKs) to encrypt data for a transfer run. You can use a CMEK to support transfers from Cloud Storage.

When you specify a CMEK with a transfer, the BigQuery Data Transfer Service applies the CMEK to any intermediate on-disk cache of ingested data so that the entire data transfer workflow is CMEK compliant.

You cannot update an existing transfer to add a CMEK if the transfer was not originally created with a CMEK. For example, you cannot change a destination table that was originally default encrypted to now be encrypted with CMEK. Conversely, you also cannot change a CMEK-encrypted destination table to have a different type of encryption.

You can update a CMEK for a transfer if the transfer configuration was originally created with a CMEK encryption. When you update a CMEK for a transfer configuration, the BigQuery Data Transfer Service propagates the CMEK to the destination tables at the next run of the transfer, where the BigQuery Data Transfer Service replaces any outdated CMEKs with the new CMEK during the transfer run. For more information, see Update a transfer.

You can also use project default keys. When you specify a project default key with a transfer, the BigQuery Data Transfer Service uses the project default key as the default key for any new transfer configurations.

Manually trigger a transfer

In addition to automatically scheduled transfers from Cloud Storage, you can manually trigger a transfer to load additional data files.

If the transfer configuration is runtime parameterized, you will need to specify a range of dates for which additional transfers will be started.

To trigger a transfer:

Console

  1. Go to the BigQuery page in the Google Cloud console.

    Go to the BigQuery page

  2. Click Data transfers.

  3. Select your transfer from the list.

  4. Click Run transfer now or Schedule backfill (for runtime parameterized transfer configurations).

    • If you clicked Run transfer now, select Run one time transfer or Run for specific date as applicable. If you selected Run for specific date, select a specific date and time:

      Run transfer now

    • If you clicked Schedule backfill, select Run one time transfer or Run for a date range as applicable. If you selected Run for a date range, select a start and end date and time:

      Schedule backfill

  5. Click Ok.

bq

Enter the bq mk command and supply the --transfer_run flag. You may either use the --run_time flag or the --start_time and --end_time flags.

bq mk \
--transfer_run \
--start_time='START_TIME' \
--end_time='END_TIME' \
RESOURCE_NAME
bq mk \
--transfer_run \
--run_time='RUN_TIME' \
RESOURCE_NAME

Where:

  • START_TIME and END_TIME are timestamps that end in Z or contain a valid time zone offset. For example:

    • 2017-08-19T12:11:35.00Z
    • 2017-05-25T00:00:00+00:00
  • RUN_TIME is a timestamp that specifies the time to schedule the data transfer run. If you want to run a one-time transfer for the current time, you may use the --run_time flag.

  • RESOURCE_NAME is the transfer's resource name (also referred to as the transfer configuration), for example, projects/myproject/locations/us/transferConfigs/1234a123-1234-1a23-1be9-12ab3c456de7. If you do not know the transfer's resource name, run the bq ls --transfer_config --transfer_location=LOCATION command to find the resource name.

API

Use the projects.locations.transferConfigs.startManualRuns method and provide the transfer configuration resource using the parent parameter.

What's next