Export table data to Cloud Storage

This page describes how to export or extract data from BigQuery tables to Cloud Storage.

After you've loaded your data into BigQuery, you can export the data in several formats. BigQuery can export up to 1 GB of logical data size to a single file. If you are exporting more than 1 GB of data, you must export your data to multiple files. When you export your data to multiple files, the size of the files will vary.

You can also export the results of a query by using the EXPORT DATA statement. You can use EXPORT DATA OPTIONS to specify the format of the exported data.

Finally, you can use a service such as Dataflow to read data from BigQuery instead of exporting it from BigLake. For more information about using Dataflow to read from and write to BigQuery, see BigQuery I/O documentation.

Export limitations

When you export data from BigQuery, note the following:

• You cannot export table data to a local file, to Google Sheets, or to Google Drive. The only supported export location is Cloud Storage. For information on saving query results, see Downloading and saving query results.
• You can export up to 1 GB of logical table data size to a single file. If you are exporting more than 1 GB of data, use a wildcard to export the data into multiple files. When you export data to multiple files, the size of the files will vary. To limit the exported file size, you can partition your data and export each partition.
• The generated file size when using the EXPORT DATA statement is not guaranteed.
• The number of files generated by an extract job can vary.
• You cannot export nested and repeated data in CSV format. Nested and repeated data are supported for Avro, JSON, and Parquet exports.
• When you export data in JSON format, INT64 (integer) data types are encoded as JSON strings to preserve 64-bit precision when the data is read by other systems.
• You cannot export data from multiple tables in a single extract job.
• You cannot choose a compression type other than GZIP when you export data using the Google Cloud console.
• When you export a table in JSON format, the symbols <, >, and & are converted by using the unicode notation \uNNNN, where N is a hexadecimal digit. For example, profit&loss becomes profit\u0026loss. This unicode conversion is done to avoid security vulnerabilities.
• The order of exported table data is not guaranteed unless you use the EXPORT DATA statement and specify an ORDER BY clause in the query_statement. When you export to multiple files, the ordered sequence is preserved globally across all generated files.
• BigQuery doesn't support Cloud Storage resource paths that include multiple consecutive slashes after the initial double slash. Cloud Storage object names can contain multiple consecutive slash ("/") characters. However, BigQuery converts multiple consecutive slashes into a single slash. For example, the following resource path, though valid in Cloud Storage, doesn't work in BigQuery: gs://bucket/my//object//name.
• Any new data loaded into BigQuery while an extract job is running won't be included in that extract job. You must create a new extract job to export the new data.

Before you begin

Grant Identity and Access Management (IAM) roles that give users the necessary permissions to perform each task in this document.

Required permissions

To perform the tasks in this document, you need the following permissions.

Permissions to export data from a BigQuery table

To export data from a BigQuery table, you need the bigquery.tables.export IAM permission.

Each of the following predefined IAM roles includes the bigquery.tables.export permission:

• roles/bigquery.dataViewer
• roles/bigquery.dataOwner
• roles/bigquery.dataEditor
• roles/bigquery.admin

Permissions to run an extract job

To run an extract job, you need the bigquery.jobs.create IAM permission.

Each of the following predefined IAM roles includes the permissions that you need in order to run an extract job:

• roles/bigquery.user
• roles/bigquery.jobUser
• roles/bigquery.admin

Permissions to write the data to the Cloud Storage bucket

To write the data to an existing Cloud Storage bucket, you need the following IAM permissions:

• storage.objects.create
• storage.objects.delete

Each of the following predefined IAM roles includes the permissions that you need in order to write the data to an existing Cloud Storage bucket:

• roles/storage.objectAdmin
• roles/storage.admin

For more information about IAM roles and permissions in BigQuery, see Predefined roles and permissions.

Export formats and compression types

BigQuery supports the following data formats and compression types for exported data.

Export data

The following sections show you how to export your table data, table metadata, and query results to Cloud Storage.

Export table data

You can export table data by:

• Using the Google Cloud console
• Using the bq extract command in the bq command-line tool
• Submitting an extract job using the API or client libraries

Select one of the following:

bq extract --location=location \
--destination_format format \
--compression compression_type \
--field_delimiter delimiter \
--print_header=boolean \
project_id:dataset.table \
gs://bucket/filename.ext

bq extract \
--compression GZIP \
'mydataset.mytable' \
gs://example-bucket/myfile.csv

bq extract \
--destination_format NEWLINE_DELIMITED_JSON \
'mydataset.mytable' \
gs://example-bucket/myfile.json

bq extract \
--destination_format AVRO \
--compression SNAPPY \
'mydataset.mytable' \
gs://example-bucket/myfile.avro

bq extract \
--destination_format CSV \
'mydataset.my_partitioned_table$0' \
gs://example-bucket/single_partition.csv

using Google.Cloud.BigQuery.V2;
using System;

public class BigQueryExtractTable
{
    public void ExtractTable(
        string projectId = "your-project-id",
        string bucketName = "your-bucket-name")
    {
        BigQueryClient client = BigQueryClient.Create(projectId);
        // Define a destination URI. Use a single wildcard URI if you think
        // your exported data will be larger than the 1 GB maximum value.
        string destinationUri = $"gs://{bucketName}/shakespeare-*.csv";
        BigQueryJob job = client.CreateExtractJob(
            projectId: "bigquery-public-data",
            datasetId: "samples",
            tableId: "shakespeare",
            destinationUri: destinationUri
        );
        job = job.PollUntilCompleted().ThrowOnAnyError();  // Waits for the job to complete.
        Console.Write($"Exported table to {destinationUri}.");
    }
}

import (
	"context"
	"fmt"

	"cloud.google.com/go/bigquery"
)

// exportTableAsCompressedCSV demonstrates using an export job to
// write the contents of a table into Cloud Storage as CSV.
func exportTableAsCSV(projectID, gcsURI string) error {
	// projectID := "my-project-id"
	// gcsUri := "gs://mybucket/shakespeare.csv"
	ctx := context.Background()
	client, err := bigquery.NewClient(ctx, projectID)
	if err != nil {
		return fmt.Errorf("bigquery.NewClient: %v", err)
	}
	defer client.Close()

	srcProject := "bigquery-public-data"
	srcDataset := "samples"
	srcTable := "shakespeare"

	gcsRef := bigquery.NewGCSReference(gcsURI)
	gcsRef.FieldDelimiter = ","

	extractor := client.DatasetInProject(srcProject, srcDataset).Table(srcTable).ExtractorTo(gcsRef)
	extractor.DisableHeader = true
	// You can choose to run the job in a specific location for more complex data locality scenarios.
	// Ex: In this example, source dataset and GCS bucket are in the US.
	extractor.Location = "US"

	job, err := extractor.Run(ctx)
	if err != nil {
		return err
	}
	status, err := job.Wait(ctx)
	if err != nil {
		return err
	}
	if err := status.Err(); err != nil {
		return err
	}
	return nil
}

import com.google.cloud.RetryOption;
import com.google.cloud.bigquery.BigQuery;
import com.google.cloud.bigquery.BigQueryException;
import com.google.cloud.bigquery.BigQueryOptions;
import com.google.cloud.bigquery.Job;
import com.google.cloud.bigquery.Table;
import com.google.cloud.bigquery.TableId;
import org.threeten.bp.Duration;

public class ExtractTableToCsv {

  public static void runExtractTableToCsv() {
    // TODO(developer): Replace these variables before running the sample.
    String projectId = "bigquery-public-data";
    String datasetName = "samples";
    String tableName = "shakespeare";
    String bucketName = "my-bucket";
    String destinationUri = "gs://" + bucketName + "/path/to/file";
    // For more information on export formats available see:
    // https://cloud.google.com/bigquery/docs/exporting-data#export_formats_and_compression_types
    // For more information on Job see:
    // https://googleapis.dev/java/google-cloud-clients/latest/index.html?com/google/cloud/bigquery/package-summary.html

    String dataFormat = "CSV";
    extractTableToCsv(projectId, datasetName, tableName, destinationUri, dataFormat);
  }

  // Exports datasetName:tableName to destinationUri as raw CSV
  public static void extractTableToCsv(
      String projectId,
      String datasetName,
      String tableName,
      String destinationUri,
      String dataFormat) {
    try {
      // Initialize client that will be used to send requests. This client only needs to be created
      // once, and can be reused for multiple requests.
      BigQuery bigquery = BigQueryOptions.getDefaultInstance().getService();

      TableId tableId = TableId.of(projectId, datasetName, tableName);
      Table table = bigquery.getTable(tableId);

      Job job = table.extract(dataFormat, destinationUri);

      // Blocks until this job completes its execution, either failing or succeeding.
      Job completedJob =
          job.waitFor(
              RetryOption.initialRetryDelay(Duration.ofSeconds(1)),
              RetryOption.totalTimeout(Duration.ofMinutes(3)));
      if (completedJob == null) {
        System.out.println("Job not executed since it no longer exists.");
        return;
      } else if (completedJob.getStatus().getError() != null) {
        System.out.println(
            "BigQuery was unable to extract due to an error: \n" + job.getStatus().getError());
        return;
      }
      System.out.println(
          "Table export successful. Check in GCS bucket for the " + dataFormat + " file.");
    } catch (BigQueryException | InterruptedException e) {
      System.out.println("Table extraction job was interrupted. \n" + e.toString());
    }
  }
}

// Import the Google Cloud client libraries
const {BigQuery} = require('@google-cloud/bigquery');
const {Storage} = require('@google-cloud/storage');

const bigquery = new BigQuery();
const storage = new Storage();

async function extractTableToGCS() {
  // Exports my_dataset:my_table to gcs://my-bucket/my-file as raw CSV.

  /**
   * TODO(developer): Uncomment the following lines before running the sample.
   */
  // const datasetId = "my_dataset";
  // const tableId = "my_table";
  // const bucketName = "my-bucket";
  // const filename = "file.csv";

  // Location must match that of the source table.
  const options = {
    location: 'US',
  };

  // Export data from the table into a Google Cloud Storage file
  const [job] = await bigquery
    .dataset(datasetId)
    .table(tableId)
    .extract(storage.bucket(bucketName).file(filename), options);

  console.log(`Job ${job.id} created.`);

  // Check the job's status for errors
  const errors = job.status.errors;
  if (errors && errors.length > 0) {
    throw errors;
  }
}

use Google\Cloud\BigQuery\BigQueryClient;

/**
 * Extracts the given table as json to given GCS bucket.
 *
 * @param string $projectId The project Id of your Google Cloud Project.
 * @param string $datasetId The BigQuery dataset ID.
 * @param string $tableId The BigQuery table ID.
 * @param string $bucketName Bucket name in Google Cloud Storage
 */
function extract_table(
    string $projectId,
    string $datasetId,
    string $tableId,
    string $bucketName
): void {
    $bigQuery = new BigQueryClient([
      'projectId' => $projectId,
    ]);
    $dataset = $bigQuery->dataset($datasetId);
    $table = $dataset->table($tableId);
    $destinationUri = "gs://{$bucketName}/{$tableId}.json";
    // Define the format to use. If the format is not specified, 'CSV' will be used.
    $format = 'NEWLINE_DELIMITED_JSON';
    // Create the extract job
    $extractConfig = $table->extract($destinationUri)->destinationFormat($format);
    // Run the job
    $job = $table->runJob($extractConfig);  // Waits for the job to complete
    printf('Exported %s to %s' . PHP_EOL, $table->id(), $destinationUri);
}

# from google.cloud import bigquery
# client = bigquery.Client()
# bucket_name = 'my-bucket'
project = "bigquery-public-data"
dataset_id = "samples"
table_id = "shakespeare"

destination_uri = "gs://{}/{}".format(bucket_name, "shakespeare.csv")
dataset_ref = bigquery.DatasetReference(project, dataset_id)
table_ref = dataset_ref.table(table_id)

extract_job = client.extract_table(
    table_ref,
    destination_uri,
    # Location must match that of the source table.
    location="US",
)  # API request
extract_job.result()  # Waits for job to complete.

print(
    "Exported {}:{}.{} to {}".format(project, dataset_id, table_id, destination_uri)
)

require "google/cloud/bigquery"

def extract_table bucket_name = "my-bucket",
                  dataset_id  = "my_dataset_id",
                  table_id    = "my_table_id"
  bigquery = Google::Cloud::Bigquery.new
  dataset  = bigquery.dataset dataset_id
  table    = dataset.table    table_id

  # Define a destination URI. Use a single wildcard URI if you think
  # your exported data will be larger than the 1 GB maximum value.
  destination_uri = "gs://#{bucket_name}/output-*.csv"

  extract_job = table.extract_job destination_uri do |config|
    # Location must match that of the source table.
    config.location = "US"
  end
  extract_job.wait_until_done! # Waits for the job to complete

  puts "Exported #{table.id} to #{destination_uri}"
end

Export table metadata

To export table metadata from Iceberg tables, use the following SQL statement:

EXPORT TABLE METADATA FROM `[[PROJECT_NAME.]DATASET_NAME.]TABLE_NAME`;

Replace the following:

• PROJECT_NAME: the name of the project for the table. The value defaults to the project that runs this query.
• DATASET_NAME: the name of the dataset for the table.
• TABLE_NAME: the name of the table.

The exported metadata is located in the STORAGE_URI/metadata folder, where STORAGE_URI is the table's storage location set in the options.

Export query results

You can export your query results to Cloud Storage in the Google Cloud console with the following steps:

1. Open the BigQuery page in the Google Cloud console.

   Go to the BigQuery page

2. Click add_box SQL query.

3. Enter a valid GoogleSQL query in the Query editor text area.

4. Click Run.

5. When the results are returned, click Save results > Cloud Storage.

6. In the Export to Google Cloud Storage dialog:

   • For GCS Location, browse for the bucket, folder, or file where you want to export the data.
   • For Export format, choose the format for your exported data: CSV, JSON (Newline Delimited), Avro, or Parquet.
   • For Compression, select a compression format or select None for no compression.

7. Click Save to export the query results.

To check on the progress of the job, expand the Job history pane and look for the job with the EXTRACT type.

Avro export details

BigQuery expresses Avro formatted data in the following ways:

• The resulting export files are Avro container files.
• Each BigQuery row is represented as an Avro record. Nested data is represented by nested record objects.
• REQUIRED fields are represented as the corresponding Avro types. For example, a BigQuery INTEGER type maps to an Avro LONG type.
• NULLABLE fields are represented as an Avro union of the corresponding type and "null".
• REPEATED fields are represented as Avro arrays.
• TIMESTAMP data types are represented as timestamp-micros logical type (it annotates an Avro LONG type) by default in both Extract jobs and Export Data SQL. (Caution: you can add use_avro_logical_types=False to Export Data Options to disable the logical type so it uses string type instead on timestamp column, but in Extract Jobs, it always uses the Avro logical type.)
• DATE data types are represented as date logical type (it annotates an Avro INT types) by default in Export Data SQL, but are represented as string type by default in Extract jobs. (Note: you can add use_avro_logical_types=False to Export Data Options to disable the logical type, or use the flag --use_avro_logical_types=True to enable the logical type in Extract jobs.)
• TIME data types are represented as timestamp-micro logical type (it annotates an Avro LONG types) by default in Export Data SQL, but are represented as string type by default in Extract jobs. (Note: you can add use_avro_logical_types=False to Export Data Options to disable the logical type, or use the flag --use_avro_logical_types=True to enable the logical type in Extract jobs.)
• DATETIME data types are represented as Avro STRING types (a string type with custom named logical type datetime) by default in Export Data SQL, but are represented as string type by default in Extract jobs. (Note: you can add use_avro_logical_types=False to Export Data Options to disable the logical type, or use the flag --use_avro_logical_types=True to enable logical type in Extract jobs.)
• RANGE types aren't supported in Avro export.

Parameterized NUMERIC(P[, S]) and BIGNUMERIC(P[, S]) data types transfer their precision and scale type parameters to the Avro decimal logical type.

• If you export a DATETIME type to Avro, you cannot load the Avro file directly back into the same table schema, because the converted STRING won't match the schema. As a workaround, load the file into a staging table. Then use a SQL query to cast the field to a DATETIME type and save the result to a new table. For more information, see Changing a column's data type .
• Export Data Option use_avro_logical_types and Extract Job flag --use_avro_logical_types are applied to all the logical types at the same time once specified.

The Avro format can't be used in combination with GZIP compression. To compress Avro data, use the bq command-line tool or the API and specify one of the supported compression types for Avro data: DEFLATE or SNAPPY.

Parquet export details

BigQuery converts GoogleSQL data types to the following Parquet data types:

The Parquet schema represents nested data as a group and repeated records as repeated groups. For more information about using nested and repeated data in BigQuery, see Specifying nested and repeated columns.

When you export a JSON column to Parquet format, it is exported as a STRING field. To load this Parquet file back into a JSON column in BigQuery, you must specify the schema in the load job (for example, by using the --schema flag). Auto schema detection with Parquet infers the column as a STRING type.

You can use the following workarounds for DATETIME types:

• Load the file into a staging table. Then use a SQL query to cast the field to a DATETIME and save the result to a new table. For more information, see Changing a column's data type.
• Provide a schema for the table by using the --schema flag in the load job. Define the datetime column as col:DATETIME.

The GEOGRAPHY logical type is represented with GeoParquet metadata added to the exported file(s).

Exporting data into one or more files

The destinationUris property indicates the one or more locations and filenames where BigQuery should export your files.

BigQuery supports a single wildcard operator (*) in each URI. The wildcard can appear anywhere in the filename component. Using the wildcard operator instructs BigQuery to create multiple sharded files based on the supplied pattern. The wildcard operator is replaced with a number (starting at 0), left-padded to 12 digits. For example, a URI with a wildcard at the end of the filename would create files with000000000000 appended to the first file, and 000000000001 appended to the second file, continuing in that pattern.

The following table describes several possible options for the destinationUris property:

gs://my-bucket/file-name.json

gs://my-bucket/file-name-000000000000.json
gs://my-bucket/file-name-000000000001.json
gs://my-bucket/file-name-000000000002.json
...

gs://my-bucket/000000000000
gs://my-bucket/000000000001
gs://my-bucket/000000000002
...

Limit the exported file size

When you export more than 1 GB of data in a single export, you must use a wildcard to export the data into multiple files and the size of the files varies. If you need to limit the maximum size of each exported file, one option is to randomly partition your data and then export each partition to a file:

1. Determine the number of partitions you need, which is equal to the total size of your data divided by the chosen exported file size. For example, if you have 8,000 MB of data and you want each exported file to be approximately 20 MB, then you need 400 partitions.

2. Create a new table that is partitioned and clustered by a new randomly generated column called export_id. The following example shows how to create a new processed_table from an existing table called source_table which requires n partitions to achieve the chosen file size:

   CREATE TABLE my_dataset.processed_table
   PARTITION BY RANGE_BUCKET(export_id, GENERATE_ARRAY(0, n, 1))
   CLUSTER BY export_id
   AS (
     SELECT *, CAST(FLOOR(n*RAND()) AS INT64) AS export_id
     FROM my_dataset.source_table
   );

3. For each integer i between 0 and n-1, run an EXPORT DATA statement on the following query:

   SELECT * EXCEPT(export_id)
   FROM my_dataset.processed_table
   WHERE export_id = i;

Extract compressed table

import (
	"context"
	"fmt"

	"cloud.google.com/go/bigquery"
)

// exportTableAsCompressedCSV demonstrates using an export job to
// write the contents of a table into Cloud Storage as compressed CSV.
func exportTableAsCompressedCSV(projectID, gcsURI string) error {
	// projectID := "my-project-id"
	// gcsURI := "gs://mybucket/shakespeare.csv"
	ctx := context.Background()
	client, err := bigquery.NewClient(ctx, projectID)
	if err != nil {
		return fmt.Errorf("bigquery.NewClient: %w", err)
	}
	defer client.Close()

	srcProject := "bigquery-public-data"
	srcDataset := "samples"
	srcTable := "shakespeare"

	gcsRef := bigquery.NewGCSReference(gcsURI)
	gcsRef.Compression = bigquery.Gzip

	extractor := client.DatasetInProject(srcProject, srcDataset).Table(srcTable).ExtractorTo(gcsRef)
	extractor.DisableHeader = true
	// You can choose to run the job in a specific location for more complex data locality scenarios.
	// Ex: In this example, source dataset and GCS bucket are in the US.
	extractor.Location = "US"

	job, err := extractor.Run(ctx)
	if err != nil {
		return err
	}
	status, err := job.Wait(ctx)
	if err != nil {
		return err
	}
	if err := status.Err(); err != nil {
		return err
	}
	return nil
}

import com.google.cloud.bigquery.BigQuery;
import com.google.cloud.bigquery.BigQueryException;
import com.google.cloud.bigquery.BigQueryOptions;
import com.google.cloud.bigquery.ExtractJobConfiguration;
import com.google.cloud.bigquery.Job;
import com.google.cloud.bigquery.JobInfo;
import com.google.cloud.bigquery.TableId;

// Sample to extract a compressed table
public class ExtractTableCompressed {

  public static void main(String[] args) {
    // TODO(developer): Replace these variables before running the sample.
    String projectName = "MY_PROJECT_NAME";
    String datasetName = "MY_DATASET_NAME";
    String tableName = "MY_TABLE_NAME";
    String bucketName = "MY-BUCKET-NAME";
    String destinationUri = "gs://" + bucketName + "/path/to/file";
    // For more information on export formats available see:
    // https://cloud.google.com/bigquery/docs/exporting-data#export_formats_and_compression_types
    String compressed = "gzip";
    // For more information on Job see:
    // https://googleapis.dev/java/google-cloud-clients/latest/index.html?com/google/cloud/bigquery/package-summary.html
    String dataFormat = "CSV";

    extractTableCompressed(
        projectName, datasetName, tableName, destinationUri, dataFormat, compressed);
  }

  public static void extractTableCompressed(
      String projectName,
      String datasetName,
      String tableName,
      String destinationUri,
      String dataFormat,
      String compressed) {
    try {
      // Initialize client that will be used to send requests. This client only needs to be created
      // once, and can be reused for multiple requests.
      BigQuery bigquery = BigQueryOptions.getDefaultInstance().getService();

      TableId tableId = TableId.of(projectName, datasetName, tableName);

      ExtractJobConfiguration extractConfig =
          ExtractJobConfiguration.newBuilder(tableId, destinationUri)
              .setCompression(compressed)
              .setFormat(dataFormat)
              .build();

      Job job = bigquery.create(JobInfo.of(extractConfig));

      // Blocks until this job completes its execution, either failing or succeeding.
      Job completedJob = job.waitFor();
      if (completedJob == null) {
        System.out.println("Job not executed since it no longer exists.");
        return;
      } else if (completedJob.getStatus().getError() != null) {
        System.out.println(
            "BigQuery was unable to extract due to an error: \n" + job.getStatus().getError());
        return;
      }
      System.out.println("Table extract compressed successful");
    } catch (BigQueryException | InterruptedException e) {
      System.out.println("Table extraction job was interrupted. \n" + e.toString());
    }
  }
}

// Import the Google Cloud client libraries
const {BigQuery} = require('@google-cloud/bigquery');
const {Storage} = require('@google-cloud/storage');

const bigquery = new BigQuery();
const storage = new Storage();

async function extractTableCompressed() {
  // Exports my_dataset:my_table to gcs://my-bucket/my-file as a compressed file.

  /**
   * TODO(developer): Uncomment the following lines before running the sample.
   */
  // const datasetId = "my_dataset";
  // const tableId = "my_table";
  // const bucketName = "my-bucket";
  // const filename = "file.csv";

  // Location must match that of the source table.
  const options = {
    location: 'US',
    gzip: true,
  };

  // Export data from the table into a Google Cloud Storage file
  const [job] = await bigquery
    .dataset(datasetId)
    .table(tableId)
    .extract(storage.bucket(bucketName).file(filename), options);

  console.log(`Job ${job.id} created.`);

  // Check the job's status for errors
  const errors = job.status.errors;
  if (errors && errors.length > 0) {
    throw errors;
  }
}

# from google.cloud import bigquery
# client = bigquery.Client()
# bucket_name = 'my-bucket'

destination_uri = "gs://{}/{}".format(bucket_name, "shakespeare.csv.gz")
dataset_ref = bigquery.DatasetReference(project, dataset_id)
table_ref = dataset_ref.table("shakespeare")
job_config = bigquery.job.ExtractJobConfig()
job_config.compression = bigquery.Compression.GZIP

extract_job = client.extract_table(
    table_ref,
    destination_uri,
    # Location must match that of the source table.
    location="US",
    job_config=job_config,
)  # API request
extract_job.result()  # Waits for job to complete.

Example use case

This example shows how you can export data to Cloud Storage.

Suppose you are streaming data to Cloud Storage from endpoint logs continuously. A daily snapshot is to be exported to Cloud Storage for backup and archival purposes. The best choice is an extract job subject to certain quotas and limitations.

Submit an extract job with the API or client libraries, passing in a unique ID as jobReference.jobId. Extract Jobs are asynchronous. Check the job status using the unique job ID used to create the job. The job completed successfully if status.status is DONE. If status.errorResult is present, the job failed and needs to be retried.

Batch data processing

Suppose a nightly batch job is used to load data by a fixed deadline. After this load job completes, a table with statistics is materialized from a query as described in the preceding section. Data from this table is retrieved and compiled into a PDF report and sent to a regulator.

Since the amount of data that needs to be read is small, use the tabledata.list API to retrieve all rows of the table in JSON dictionary format. If there is more than one page of data, the results have the pageToken property set. To retrieve the next page of results, make another tabledata.list call and include the token value as the pageToken parameter. If the API call fails with a 5xx error, retry with exponential backoff. Most 4xx errors cannot be retried. For better decoupling of BigQuery export and report generation, results should be persisted to disk.

Quota policy

For information on extract job quotas, see Extract jobs on the Quotas and limits page.

Usage for extract jobs are available in the INFORMATION_SCHEMA. The job entry in the JOBS_BY_* system tables for the extract job contains a total_bytes_processed value that can be used to monitor the aggregate usage to ensure that it stays under 50 TiB per-day. To learn how to query the INFORMATION_SCHEMA.JOBS view to get the total_bytes_processed value, see INFORMATION_SCHEMA.JOBS schema

View current quota usage

You can view your current usage of query, load, extract, or copy jobs by running an INFORMATION_SCHEMA query to view metadata about the jobs ran over a specified time period. You can compare your current usage against the quota limit to determine your quota usage for a particular type of job. The following example query uses the INFORMATION_SCHEMA.JOBS view to list the number of query, load, extract, and copy jobs by project:

SELECT
  sum(case  when job_type="QUERY" then 1 else 0 end) as QRY_CNT,
  sum(case  when job_type="LOAD" then 1 else 0 end) as LOAD_CNT,
  sum(case  when job_type="EXTRACT" then 1 else 0 end) as EXT_CNT,
  sum(case  when job_type="COPY" then 1 else 0 end) as CPY_CNT
FROM `region-REGION_NAME`.INFORMATION_SCHEMA.JOBS_BY_PROJECT
WHERE date(creation_time)= CURRENT_DATE()

You can set up a Cloud Monitoring alerting policy that monitors the number of bytes exported.

1. In the Google Cloud console, go to the notifications Alerting page:

   Go to Alerting

   If you use the search bar to find this page, then select the result whose subheading is Monitoring.

2. On the Alerting page, click Create Policy.

3. Under Policy configuration mode, select Code editor (MQL or PromQL).

4. In the PromQL query editor, enter the following query:

   (
     sum by (project_id, quota_metric, location) (increase({"serviceruntime.googleapis.com/quota/rate/net_usage", monitored_resource="consumer_quota", service="bigquery.googleapis.com"}[1m]))
     /
     max by (project_id, quota_metric, location) ({"serviceruntime.googleapis.com/quota/limit", monitored_resource="consumer_quota", service="bigquery.googleapis.com", limit_name="ExtractBytesPerDay"})
   ) > 0.01
   

   If Auto-run isn't enabled, then click Run Query.

5. Configure the rest of your alert and click Create policy.

For detailed steps about creating PromQL-based alerting policies, see Create PromQL-based alerting policies (Console).

Troubleshooting

Diagnose and troubleshoot issues with extract jobs.

Diagnose issues using Logs Explorer

To diagnose issues with extract jobs, you can use the Logs Explorer to review the logs for a specific extract job and identify possible errors. The following Logs Explorer filter returns information about your extract jobs:

resource.type="bigquery_resource"
protoPayload.methodName="jobservice.insert"
(protoPayload.serviceData.jobInsertRequest.resource.jobConfiguration.query.query=~"EXPORT" OR
protoPayload.serviceData.jobCompletedEvent.eventName="extract_job_completed" OR
protoPayload.serviceData.jobCompletedEvent.job.jobConfiguration.query.query=~"EXPORT")

Exceeded extract bytes per day quota error

BigQuery returns this error when the extraction exceeds the default 50 TiB daily limit in a project. For more information about extract job limits, see Extract jobs.

Error message

Your usage exceeded quota for ExtractBytesPerDay

Diagnosis

If you are exporting a table that is larger than 50 TiB, the export fails because it exceeds the extraction limit. If you want to export table data for specific table partitions, you can use a partition decorator to identify the partitions to export.

If you would like to gather usages of exports data over recent days, you can try the following:

• View the quotas for your project with filter criteria such as Name: Extract bytes per day or Metric: bigquery.googleapis.com/quota/extract/bytes along with the Show usage chart to see your usage trend over a few days.

• Alternatively you can query INFORMATION_SCHEMA.JOBS_BY_PROJECT to see your total extract bytes over a few days. For example, the following query returns the daily total bytes processed by EXTRACT jobs in the past seven days.

  SELECT
  TIMESTAMP_TRUNC(creation_time, DAY) AS day,
  SUM ( total_bytes_processed ) / POW(1024, 3) AS total_gibibytes_processed
  FROM
  `region-REGION_NAME`.INFORMATION_SCHEMA.JOBS_BY_PROJECT
  WHERE
  creation_time BETWEEN TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL 7 DAY) AND CURRENT_TIMESTAMP()
  AND job_type = "EXTRACT"
  GROUP BY 1
  ORDER BY 2 DESC

• You can then further refine the results by identifying the specific jobs that are consuming more bytes than expected. The following example returns the top 100 EXTRACT jobs which are consuming more than 100 GB processed over the past seven days.

  SELECT
  creation_time,
  job_id,
  total_bytes_processed/POW(1024, 3) AS total_gigabytes_processed
  FROM
  `region-REGION_NAME`.INFORMATION_SCHEMA.JOBS_BY_PROJECT
  WHERE
  creation_time BETWEEN TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL 7 DAY) AND CURRENT_TIMESTAMP()
  AND job_type="EXTRACT"
  AND total_bytes_processed > (POW(1024, 3) * 100)
  ORDER BY
  total_bytes_processed DESC
  LIMIT 100

You can alternatively use the jobs explorer with filters like Bytes processed more than to filter for high processing jobs for a specified period of time.

Resolution

One method of resolving this quota error is to create a slot reservation and assign your project into the reservation with the PIPELINE job type. This method can bypass the limit check since it uses your dedicated reservations rather than a free shared slot pool. If needed, the reservation can be deleted if you want to use a shared slot pool later on.

For alternative approaches that allow exporting more than 50 TiB, see the notes section in Extract jobs.

Failed to export on nested schema error

This error occurs when you try to export a table containing nested or repeated fields, such as STRUCT or ARRAY data types, to a flat-file format like CSV or newline-delimited JSON.

Error message

Failed to export: Operation cannot be performed on a nested schema. Field: [COLUMN_NAME]

This error occurs because flat-file formats like CSV can't represent hierarchical or multi-valued data. BigQuery exports require you to flatten or transform this complex data into a tabular structure before it can be written in a flat-file format. In this case, the COLUMN_NAME refers to a column with nested or repeated fields that need to be flattened or transformed before the data can be exported.

Resolution

Depending on your source schema, there are several methods that you can use to flatten your schema to prepare it for export.

The following example shows a schema with nested STRUCT fields and a query that flattens them for export. In the example, the export query fails due to the nested fields in the user_details column. You can use the suggested fix to flatten the file structure into two columns, user_details.name and user_details.email.

  -- Schema example:
  root
     id (INTEGER)
     user_details (RECORD / STRUCT, NULLABLE)
        name (STRING)
        email (STRING)
     created_at (TIMESTAMP)

  -- Export query:
  SELECT * FROM my_dataset.my_table;

  -- Suggested fix:
  SELECT
      id,
      user_details.name AS user_name,
      user_details.email AS user_email,
      created_at
  FROM
      my_dataset.my_table;

The following example shows a schema with repeated ARRAY fields and a query that flattens them for export. In the example, the export query fails due to the repeated fields in the line_items column. You can use the suggested fix which uses the UNNEST() operator to convert the array elements into individual rows.

  -- Schema example:
  root
     order_id (STRING)
     customer_name (STRING)
     line_items (REPEATED / ARRAY of RECORD/STRUCT, NULLABLE)
         product_id (STRING)
         quantity (INTEGER)


  -- Export query:
  SELECT * FROM my_dataset.my_orders_table;

  -- Suggested fix:
  SELECT
      t.order_id,
      t.customer_name,
      item.product_id,
      item.quantity
  FROM
      my_dataset.my_orders_table AS t
  LEFT JOIN
      UNNEST(t.line_items) AS item
  ;

Alternatively, you can use ARRAY_TO_STRING() if you only need to represent an array, such as an ARRAY<STRING>, as a single string column in your CSV file.

ARRAY_TO_STRING() concatenates all elements of the interests array into a single string, using a comma as a separator. This transformation is useful when the exact structure of each array element isn't crucial for the export.

  -- Schema example:
  root
     user_id (STRING)
     interests (REPEATED / ARRAY of STRING, NULLABLE)

  -- Suggested fix:
  SELECT
      user_id,
      ARRAY_TO_STRING(interests, ', ') AS user_interests
  FROM
      my_dataset.my_users_table;

Pricing

For information on data export pricing, see the BigQuery pricing page.

Once the data is exported, you are charged for storing the data in Cloud Storage. For more information, see Cloud Storage pricing.

Table security

To control access to tables in BigQuery, see Control access to resources with IAM.

What's next

• To learn more about the Google Cloud console, see Using the Google Cloud console.
• To learn more about the bq command-line tool, see Using the bq command-line tool.
• To learn how to create an application using the BigQuery API client libraries, see Client library quickstart.