Pub/Sub Avro to BigQuery template

The Pub/Sub Avro to BigQuery template is a streaming pipeline that ingests Avro data from a Pub/Sub subscription into a BigQuery table. Any errors which occur while writing to the BigQuery table are streamed into a Pub/Sub unprocessed topic.

Pipeline requirements

• The input Pub/Sub subscription must exist.
• The schema file for the Avro records must exist on Cloud Storage.
• The unprocessed Pub/Sub topic must exist.
• The output BigQuery dataset must exist.

Template parameters

Required parameters

• schemaPath : The Cloud Storage location of the Avro schema file. For example, gs://path/to/my/schema.avsc.
• inputSubscription : The Pub/Sub input subscription to read from. (Example: projects/<PROJECT_ID>/subscription/<SUBSCRIPTION_ID>).
• outputTableSpec : The BigQuery output table location to write the output to. For example, <PROJECT_ID>:<DATASET_NAME>.<TABLE_NAME>.Depending on the createDisposition specified, the output table might be created automatically using the user provided Avro schema.
• outputTopic : The Pub/Sub topic to use for unprocessed records. (Example: projects/<PROJECT_ID>/topics/<TOPIC_NAME>).

Optional parameters

• useStorageWriteApiAtLeastOnce : When using the Storage Write API, specifies the write semantics. To use at-least-once semantics (https://beam.apache.org/documentation/io/built-in/google-bigquery/#at-least-once-semantics), set this parameter to true. To use exactly-once semantics, set the parameter to false. This parameter applies only when useStorageWriteApi is true. The default value is false.
• writeDisposition : The BigQuery WriteDisposition (https://cloud.google.com/bigquery/docs/reference/rest/v2/Job#jobconfigurationload) value. For example, WRITE_APPEND, WRITE_EMPTY, or WRITE_TRUNCATE. Defaults to WRITE_APPEND.
• createDisposition : The BigQuery CreateDisposition (https://cloud.google.com/bigquery/docs/reference/rest/v2/Job#jobconfigurationload). For example, CREATE_IF_NEEDED and CREATE_NEVER. Defaults to CREATE_IF_NEEDED.
• useStorageWriteApi : If true, the pipeline uses the BigQuery Storage Write API (https://cloud.google.com/bigquery/docs/write-api). The default value is false. For more information, see Using the Storage Write API (https://beam.apache.org/documentation/io/built-in/google-bigquery/#storage-write-api).
• numStorageWriteApiStreams : When using the Storage Write API, specifies the number of write streams. If useStorageWriteApi is true and useStorageWriteApiAtLeastOnce is false, then you must set this parameter. Defaults to: 0.
• storageWriteApiTriggeringFrequencySec : When using the Storage Write API, specifies the triggering frequency, in seconds. If useStorageWriteApi is true and useStorageWriteApiAtLeastOnce is false, then you must set this parameter.

Run the template

Console

1. Go to the Dataflow Create job from template page.
Go to Create job from template
2. In the Job name field, enter a unique job name.
3. Optional: For Regional endpoint, select a value from the drop-down menu. The default region is us-central1.

   For a list of regions where you can run a Dataflow job, see Dataflow locations.

4. From the Dataflow template drop-down menu, select the Pub/Sub Avro to BigQuery template.
5. In the provided parameter fields, enter your parameter values.
6. Click Run job.

gcloud

Note: To use the Google Cloud CLI to run flex templates, you must have Google Cloud CLI version 284.0.0 or later.

In your shell or terminal, run the template:

gcloud dataflow flex-template run JOB_NAME \
    --region=REGION_NAME \
    --template-file-gcs-location=gs://dataflow-templates-REGION_NAME/VERSION/flex/PubSub_Avro_to_BigQuery \
    --parameters \
schemaPath=SCHEMA_PATH,\
inputSubscription=SUBSCRIPTION_NAME,\
outputTableSpec=BIGQUERY_TABLE,\
outputTopic=DEADLETTER_TOPIC
  

Replace the following:

• JOB_NAME: a unique job name of your choice
• REGION_NAME: the region where you want to deploy your Dataflow job—for example, us-central1
• VERSION: the version of the template that you want to use

  You can use the following values:

  • latest to use the latest version of the template, which is available in the non-dated parent folder in the bucket— gs://dataflow-templates-REGION_NAME/latest/
  • the version name, like 2023-09-12-00_RC00, to use a specific version of the template, which can be found nested in the respective dated parent folder in the bucket— gs://dataflow-templates-REGION_NAME/
  Caution: The latest version of templates might update with breaking changes. Your production environments should use templates kept in the most recent dated parent folder to prevent these breaking changes from affecting your production workflows.
• SCHEMA_PATH: the Cloud Storage path to the Avro schema file (for example, gs://MyBucket/file.avsc)
• SUBSCRIPTION_NAME: the Pub/Sub input subscription name
• BIGQUERY_TABLE: the BigQuery output table name
• DEADLETTER_TOPIC: the Pub/Sub topic to use for the unprocessed queue

API

To run the template using the REST API, send an HTTP POST request. For more information on the API and its authorization scopes, see projects.templates.launch.

POST https://dataflow.googleapis.com/v1b3/projects/PROJECT_ID/locations/LOCATION/flexTemplates:launch
{
   "launch_parameter": {
      "jobName": "JOB_NAME",
      "containerSpecGcsPath": "gs://dataflow-templates-LOCATION/VERSION/flex/PubSub_Avro_to_BigQuery",
      "parameters": {
          "schemaPath": "SCHEMA_PATH",
          "inputSubscription": "SUBSCRIPTION_NAME",
          "outputTableSpec": "BIGQUERY_TABLE",
          "outputTopic": "DEADLETTER_TOPIC"
      }
   }
}
  

Replace the following:

• JOB_NAME: a unique job name of your choice
• LOCATION: the region where you want to deploy your Dataflow job—for example, us-central1
• VERSION: the version of the template that you want to use

  You can use the following values:

  • latest to use the latest version of the template, which is available in the non-dated parent folder in the bucket— gs://dataflow-templates-REGION_NAME/latest/
  • the version name, like 2023-09-12-00_RC00, to use a specific version of the template, which can be found nested in the respective dated parent folder in the bucket— gs://dataflow-templates-REGION_NAME/
  Caution: The latest version of templates might update with breaking changes. Your production environments should use templates kept in the most recent dated parent folder to prevent these breaking changes from affecting your production workflows.
• SCHEMA_PATH: the Cloud Storage path to the Avro schema file (for example, gs://MyBucket/file.avsc)
• SUBSCRIPTION_NAME: the Pub/Sub input subscription name
• BIGQUERY_TABLE: the BigQuery output table name
• DEADLETTER_TOPIC: the Pub/Sub topic to use for the unprocessed queue

Template source code

Java

/*
 * Copyright (C) 2020 Google LLC
 *
 * Licensed under the Apache License, Version 2.0 (the "License"); you may not
 * use this file except in compliance with the License. You may obtain a copy of
 * the License at
 *
 *   http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
 * License for the specific language governing permissions and limitations under
 * the License.
 */
package com.google.cloud.teleport.v2.templates;

import com.google.cloud.teleport.metadata.Template;
import com.google.cloud.teleport.metadata.TemplateCategory;
import com.google.cloud.teleport.metadata.TemplateParameter;
import com.google.cloud.teleport.v2.common.UncaughtExceptionLogger;
import com.google.cloud.teleport.v2.options.BigQueryCommonOptions.WriteOptions;
import com.google.cloud.teleport.v2.options.BigQueryStorageApiStreamingOptions;
import com.google.cloud.teleport.v2.options.PubsubCommonOptions.ReadSubscriptionOptions;
import com.google.cloud.teleport.v2.options.PubsubCommonOptions.WriteTopicOptions;
import com.google.cloud.teleport.v2.templates.PubsubAvroToBigQuery.PubsubAvroToBigQueryOptions;
import com.google.cloud.teleport.v2.transforms.BigQueryConverters;
import com.google.cloud.teleport.v2.transforms.ErrorConverters;
import com.google.cloud.teleport.v2.utils.BigQueryIOUtils;
import com.google.cloud.teleport.v2.utils.SchemaUtils;
import org.apache.avro.Schema;
import org.apache.avro.generic.GenericRecord;
import org.apache.beam.sdk.Pipeline;
import org.apache.beam.sdk.PipelineResult;
import org.apache.beam.sdk.extensions.avro.coders.AvroCoder;
import org.apache.beam.sdk.io.gcp.bigquery.WriteResult;
import org.apache.beam.sdk.io.gcp.pubsub.PubsubIO;
import org.apache.beam.sdk.options.Default;
import org.apache.beam.sdk.options.PipelineOptionsFactory;
import org.apache.beam.sdk.options.Validation.Required;
import org.apache.beam.sdk.schemas.transforms.Convert;
import org.apache.beam.sdk.values.Row;

/**
 * A Dataflow pipeline to stream <a href="https://avro.apache.org/">Apache Avro</a> records from
 * Pub/Sub into a BigQuery table.
 *
 * <p>Any persistent failures while writing to BigQuery will be written to a Pub/Sub dead-letter
 * topic.
 *
 * <p>Check out <a
 * href="https://github.com/GoogleCloudPlatform/DataflowTemplates/blob/main/v2/pubsub-binary-to-bigquery/README_PubSub_Avro_to_BigQuery.md">README</a>
 * for instructions on how to use or modify this template.
 */
@Template(
    name = "PubSub_Avro_to_BigQuery",
    category = TemplateCategory.STREAMING,
    displayName = "Pub/Sub Avro to BigQuery",
    description =
        "The Pub/Sub Avro to BigQuery template is a streaming pipeline that ingests Avro data from a Pub/Sub "
            + "subscription into a BigQuery table. Any errors which occur while writing to the BigQuery table are streamed into a Pub/Sub unprocessed topic.",
    optionsClass = PubsubAvroToBigQueryOptions.class,
    flexContainerName = "pubsub-avro-to-bigquery",
    documentation =
        "https://cloud.google.com/dataflow/docs/guides/templates/provided/pubsub-avro-to-bigquery",
    contactInformation = "https://cloud.google.com/support",
    requirements = {
      "The input Pub/Sub subscription must exist.",
      "The schema file for the Avro records must exist on Cloud Storage.",
      "The unprocessed Pub/Sub topic must exist.",
      "The output BigQuery dataset must exist."
    },
    streaming = true,
    supportsAtLeastOnce = true)
public final class PubsubAvroToBigQuery {
  /**
   * Validates input flags and executes the Dataflow pipeline.
   *
   * @param args command line arguments to the pipeline
   */
  public static void main(String[] args) {
    UncaughtExceptionLogger.register();

    PubsubAvroToBigQueryOptions options =
        PipelineOptionsFactory.fromArgs(args)
            .withValidation()
            .as(PubsubAvroToBigQueryOptions.class);

    run(options);
  }

  /**
   * Provides custom {@link org.apache.beam.sdk.options.PipelineOptions} required to execute the
   * {@link PubsubAvroToBigQuery} pipeline.
   */
  public interface PubsubAvroToBigQueryOptions
      extends ReadSubscriptionOptions,
          WriteOptions,
          WriteTopicOptions,
          BigQueryStorageApiStreamingOptions {

    @TemplateParameter.GcsReadFile(
        order = 1,
        description = "Cloud Storage path to the Avro schema file",
        helpText =
            "The Cloud Storage location of the Avro schema file. For example, `gs://path/to/my/schema.avsc`.")
    @Required
    String getSchemaPath();

    void setSchemaPath(String schemaPath);

    // Hide the UseStorageWriteApiAtLeastOnce in the UI, because it will automatically be turned
    // on when pipeline is running on ALO mode and using the Storage Write API
    @TemplateParameter.Boolean(
        order = 2,
        optional = true,
        parentName = "useStorageWriteApi",
        parentTriggerValues = {"true"},
        description = "Use at at-least-once semantics in BigQuery Storage Write API",
        helpText =
            " When using the Storage Write API, specifies the write semantics. To use"
                + " at-least-once semantics (https://beam.apache.org/documentation/io/built-in/google-bigquery/#at-least-once-semantics), set this parameter to true. To use exactly-once"
                + " semantics, set the parameter to `false`. This parameter applies only when `useStorageWriteApi` is `true`. The default value is `false`.",
        hiddenUi = true)
    @Default.Boolean(false)
    @Override
    Boolean getUseStorageWriteApiAtLeastOnce();

    void setUseStorageWriteApiAtLeastOnce(Boolean value);
  }

  /**
   * Runs the pipeline with the supplied options.
   *
   * @param options execution parameters to the pipeline
   * @return result of the pipeline execution as a {@link PipelineResult}
   */
  private static PipelineResult run(PubsubAvroToBigQueryOptions options) {
    BigQueryIOUtils.validateBQStorageApiOptionsStreaming(options);

    // Create the pipeline.
    Pipeline pipeline = Pipeline.create(options);

    Schema schema = SchemaUtils.getAvroSchema(options.getSchemaPath());

    WriteResult writeResults =
        pipeline
            .apply(
                "Read Avro records",
                PubsubIO.readAvroGenericRecords(schema)
                    .fromSubscription(options.getInputSubscription())
                    .withDeadLetterTopic(options.getOutputTopic()))
            // Workaround for BEAM-12256. Eagerly convert to rows to avoid
            // the RowToGenericRecord function that doesn't handle all data
            // types.
            // TODO: Remove this workaround when a fix for BEAM-12256 is
            // released.
            .apply(Convert.toRows())
            .apply(
                "Write to BigQuery",
                BigQueryConverters.<Row>createWriteTransform(options).useBeamSchema());

    BigQueryIOUtils.writeResultToBigQueryInsertErrors(writeResults, options)
        .apply(
            "Create error payload",
            ErrorConverters.BigQueryInsertErrorToPubsubMessage.<GenericRecord>newBuilder()
                .setPayloadCoder(AvroCoder.of(schema))
                .setTranslateFunction(BigQueryConverters.TableRowToGenericRecordFn.of(schema))
                .build())
        .apply("Write failed records", PubsubIO.writeMessages().to(options.getOutputTopic()));

    // Execute the pipeline and return the result.
    return pipeline.run();
  }
}

What's next

• Learn about Dataflow templates.
• See the list of Google-provided templates.