BigQuery to Elasticsearch template

Beta

This feature is subject to the "Pre-GA Offerings Terms" in the General Service Terms section of the Service Specific Terms. Pre-GA features are available "as is" and might have limited support. For more information, see the launch stage descriptions.

The BigQuery to Elasticsearch template is a batch pipeline that ingests data from a BigQuery table into Elasticsearch as documents. The template can either read the entire table or read specific records using a supplied query.

Pipeline requirements

• The source BigQuery table must exist.
• A Elasticsearch host on a Google Cloud instance or on Elastic Cloud with Elasticsearch version 7.0 or later. Must be accessible from the Dataflow worker machines.

Template parameters

Parameter	Description
connectionUrl	Elasticsearch URL in the format https://hostname:[port] or specify CloudID if using Elastic Cloud.
apiKey	Base64 Encoded API key used for authentication.
index	The Elasticsearch index toward which the requests will be issued, For example, my-index.
inputTableSpec	(Optional) BigQuery table to read from to insert into Elasticsearch. Either table or query must be provided. For example, projectId:datasetId.tablename.
query	(Optional) SQL query to pull data from BigQuery. Either table or query must be provided.
useLegacySql	(Optional) Set to true to use legacy SQL (only applicable if supplying query). Default: false.
batchSize	(Optional) Batch size in number of documents. Default: 1000.
batchSizeBytes	(Optional) Batch size in number of bytes. Default: 5242880 (5mb).
maxRetryAttempts	(Optional) Max retry attempts, must be > 0. Default: no retries.
maxRetryDuration	(Optional) Max retry duration in milliseconds, must be > 0. Default: no retries.
propertyAsIndex	(Optional) A property in the document being indexed whose value will specify _index metadata to be included with document in bulk request (takes precedence over an _index UDF). Default: none.
propertyAsId	(Optional) A property in the document being indexed whose value will specify _id metadata to be included with document in bulk request (takes precedence over an _id UDF). Default: none.
javaScriptIndexFnGcsPath	(Optional) The Cloud Storage path to the JavaScript UDF source for a function that will specify _index metadata to be included with document in bulk request. Default: none.
javaScriptIndexFnName	(Optional) UDF JavaScript function name for function that will specify _index metadata to be included with document in bulk request. Default: none.
javaScriptIdFnGcsPath	(Optional) The Cloud Storage path to the JavaScript UDF source for a function that will specify _id metadata to be included with document in bulk request. Default: none.
javaScriptIdFnName	(Optional) UDF JavaScript function name for function that will specify _id metadata to be included with document in bulk request. Default: none.
javaScriptTypeFnGcsPath	(Optional) The Cloud Storage path to the JavaScript UDF source for a function that will specify _type metadata to be included with document in bulk request. Default: none.
javaScriptTypeFnName	(Optional) UDF JavaScript function name for function that will specify _type metadata to be included with document in bulk request. Default: none.
javaScriptIsDeleteFnGcsPath	(Optional) The Cloud Storage path to JavaScript UDF source for function that will determine if document should be deleted rather than inserted or updated. The function should return string value "true" or "false". Default: none.
javaScriptIsDeleteFnName	(Optional) UDF JavaScript function name for function that will determine if document should be deleted rather than inserted or updated. The function should return string value "true" or "false". Default: none.
usePartialUpdate	(Optional) Whether to use partial updates (update rather than create or index, allowing partial docs) with Elasticsearch requests. Default: false.
bulkInsertMethod	(Optional) Whether to use INDEX (index, allows upserts) or CREATE (create, errors on duplicate _id) with Elasticsearch bulk requests. Default: CREATE.

User-defined functions

This template supports user-defined functions (UDFs) at several points in the pipeline, described below. For more information, see Create user-defined functions for Dataflow templates.

Index function

Returns the index to which the document belongs.

Template parameters:

• javaScriptIndexFnGcsPath: the Cloud Storage URI of the JavaScript file.
• javaScriptIndexFnName: the name of the JavaScript function.

Function specification:

• Input: the Elasticsearch document, serialized as a JSON string.
• Output: the value of the document's _index metadata field.

Document ID function

Returns the document ID.

Template parameters:

• javaScriptIdFnGcsPath: the Cloud Storage URI of the JavaScript file.
• javaScriptIdFnName: the name of the JavaScript function.

Function specification:

• Input: the Elasticsearch document, serialized as a JSON string.
• Output: the value of the document's _id metadata field.

Document deletion function

Specifies whether to delete a document. To use this function, set the bulk insert mode to INDEX and provide a document ID function.

Template parameters:

• javaScriptIsDeleteFnGcsPath: the Cloud Storage URI of the JavaScript file.
• javaScriptIsDeleteFnName: the name of the JavaScript function.

Function specification:

• Input: the Elasticsearch document, serialized as a JSON string.
• Output: return the string "true" to delete the document, or "false" to upsert the document.

Mapping type function

Returns the document's mapping type.

Caution: Mapping types were removed in Elasticsearch 8.0.

Template parameters:

• javaScriptTypeFnGcsPath: the Cloud Storage URI of the JavaScript file.
• javaScriptTypeFnName: the name of the JavaScript function.

Function specification:

• Input: the Elasticsearch document, serialized as a JSON string.
• Output: the value of the document's _type metadata field.

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 BigQuery to Elasticsearch 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 \
    --project=PROJECT_ID \
    --region=REGION_NAME \
    --template-file-gcs-location=gs://dataflow-templates-REGION_NAME/VERSION/flex/BigQuery_to_Elasticsearch \
    --parameters \
inputTableSpec=INPUT_TABLE_SPEC,\
connectionUrl=CONNECTION_URL,\
apiKey=APIKEY,\
index=INDEX

Replace the following:

• PROJECT_ID: the Google Cloud project ID where you want to run the Dataflow job
• 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.
• INPUT_TABLE_SPEC: your BigQuery table name.
• CONNECTION_URL: your Elasticsearch URL.
• APIKEY: your base64 encoded API key for authentication.
• INDEX: your Elasticsearch index.

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",
      "parameters": {
          "inputTableSpec": "INPUT_TABLE_SPEC",
          "connectionUrl": "CONNECTION_URL",
          "apiKey": "APIKEY",
          "index": "INDEX"
      },
      "containerSpecGcsPath": "gs://dataflow-templates-LOCATION/VERSION/flex/BigQuery_to_Elasticsearch",
   }
}

Replace the following:

• PROJECT_ID: the Google Cloud project ID where you want to run the Dataflow job
• 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.
• INPUT_TABLE_SPEC: your BigQuery table name.
• CONNECTION_URL: your Elasticsearch URL.
• APIKEY: your base64 encoded API key for authentication.
• INDEX: your Elasticsearch index.

Template source code

Java

/*
 * Copyright (C) 2021 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.elasticsearch.templates;

import com.google.cloud.teleport.metadata.MultiTemplate;
import com.google.cloud.teleport.metadata.Template;
import com.google.cloud.teleport.metadata.TemplateCategory;
import com.google.cloud.teleport.v2.common.UncaughtExceptionLogger;
import com.google.cloud.teleport.v2.elasticsearch.options.BigQueryToElasticsearchOptions;
import com.google.cloud.teleport.v2.elasticsearch.transforms.WriteToElasticsearch;
import com.google.cloud.teleport.v2.transforms.BigQueryConverters.ReadBigQueryTableRows;
import com.google.cloud.teleport.v2.transforms.BigQueryConverters.TableRowToJsonFn;
import com.google.cloud.teleport.v2.transforms.JavascriptTextTransformer.TransformTextViaJavascript;
import com.google.cloud.teleport.v2.transforms.PythonExternalTextTransformer;
import com.google.common.base.Strings;
import org.apache.beam.sdk.Pipeline;
import org.apache.beam.sdk.PipelineResult;
import org.apache.beam.sdk.options.PipelineOptionsFactory;
import org.apache.beam.sdk.transforms.ParDo;
import org.apache.beam.sdk.values.PCollection;

/**
 * The {@link BigQueryToElasticsearch} pipeline exports data from a BigQuery table to Elasticsearch.
 *
 * <p>Check out <a
 * href="https://github.com/GoogleCloudPlatform/DataflowTemplates/blob/main/v2/googlecloud-to-elasticsearch/README_BigQuery_to_Elasticsearch.md">README</a>
 * for instructions on how to use or modify this template.
 */
@MultiTemplate({
  @Template(
      name = "BigQuery_to_Elasticsearch",
      category = TemplateCategory.BATCH,
      displayName = "BigQuery to Elasticsearch",
      description =
          "The BigQuery to Elasticsearch template is a batch pipeline that ingests data from a BigQuery table into Elasticsearch as documents. "
              + "The template can either read the entire table or read specific records using a supplied query.",
      optionsClass = BigQueryToElasticsearchOptions.class,
      skipOptions = {
        "javascriptTextTransformReloadIntervalMinutes",
        "pythonExternalTextTransformGcsPath",
        "pythonExternalTextTransformFunctionName"
      },
      flexContainerName = "bigquery-to-elasticsearch",
      documentation =
          "https://cloud.google.com/dataflow/docs/guides/templates/provided/bigquery-to-elasticsearch",
      contactInformation = "https://cloud.google.com/support",
      preview = true,
      requirements = {
        "The source BigQuery table must exist.",
        "A Elasticsearch host on a Google Cloud instance or on Elastic Cloud with Elasticsearch version 7.0 or above and should be accessible from the Dataflow worker machines.",
      }),
  @Template(
      name = "BigQuery_to_Elasticsearch_Xlang",
      category = TemplateCategory.BATCH,
      displayName = "BigQuery to Elasticsearch with Python UDFs",
      type = Template.TemplateType.XLANG,
      description =
          "The BigQuery to Elasticsearch template is a batch pipeline that ingests data from a BigQuery table into Elasticsearch as documents. "
              + "The template can either read the entire table or read specific records using a supplied query.",
      optionsClass = BigQueryToElasticsearchOptions.class,
      skipOptions = {
        "javascriptTextTransformReloadIntervalMinutes",
        "javascriptTextTransformGcsPath",
        "javascriptTextTransformFunctionName"
      },
      flexContainerName = "bigquery-to-elasticsearch-xlang",
      documentation =
          "https://cloud.google.com/dataflow/docs/guides/templates/provided/bigquery-to-elasticsearch",
      contactInformation = "https://cloud.google.com/support",
      preview = true,
      requirements = {
        "The source BigQuery table must exist.",
        "A Elasticsearch host on a Google Cloud instance or on Elastic Cloud with Elasticsearch version 7.0 or above and should be accessible from the Dataflow worker machines.",
      })
})
public class BigQueryToElasticsearch {
  /**
   * Main entry point for pipeline execution.
   *
   * @param args Command line arguments to the pipeline.
   */
  public static void main(String[] args) {
    UncaughtExceptionLogger.register();

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

    run(options);
  }

  /**
   * Runs the pipeline with the supplied options.
   *
   * @param options The execution parameters to the pipeline.
   * @return The result of the pipeline execution.
   */
  private static PipelineResult run(BigQueryToElasticsearchOptions options) {

    // Create the pipeline.
    Pipeline pipeline = Pipeline.create(options);
    /*
     * Steps: 1) Read records from BigQuery via BigQueryIO.
     *        2) Create json string from Table Row.
     *        3) Write records to Elasticsearch.
     */

    boolean useJavascriptUdf = !Strings.isNullOrEmpty(options.getJavascriptTextTransformGcsPath());
    boolean usePythonUdf = !Strings.isNullOrEmpty(options.getPythonExternalTextTransformGcsPath());
    if (useJavascriptUdf && usePythonUdf) {
      throw new IllegalArgumentException(
          "Either javascript or Python gcs path must be provided, but not both.");
    }

    /*
     * Step #1: Read from BigQuery. If a query is provided then it is used to get the TableRows.
     */
    PCollection<String> readJsonDocuments =
        pipeline
            .apply(
                "ReadFromBigQuery",
                ReadBigQueryTableRows.newBuilder()
                    .setOptions(options.as(BigQueryToElasticsearchOptions.class))
                    .build())

            /*
             * Step #2: Convert table rows to JSON documents.
             */
            .apply("TableRowsToJsonDocument", ParDo.of(new TableRowToJsonFn()));

    /*
     * Step #3: Apply UDF functions (if specified)
     */
    PCollection<String> udfOut;
    if (usePythonUdf) {
      udfOut =
          readJsonDocuments
              .apply(
                  "MapToRecord",
                  PythonExternalTextTransformer.FailsafeRowPythonExternalUdf
                      .stringMappingFunction())
              .setRowSchema(PythonExternalTextTransformer.FailsafeRowPythonExternalUdf.ROW_SCHEMA)
              .apply(
                  "InvokeUDF",
                  PythonExternalTextTransformer.FailsafePythonExternalUdf.newBuilder()
                      .setFileSystemPath(options.getPythonExternalTextTransformGcsPath())
                      .setFunctionName(options.getPythonExternalTextTransformFunctionName())
                      .build())
              .apply(
                  "MapToStringElements",
                  ParDo.of(new PythonExternalTextTransformer.RowToStringElementFn()));
    } else {
      udfOut =
          readJsonDocuments.apply(
              TransformTextViaJavascript.newBuilder()
                  .setFileSystemPath(options.getJavascriptTextTransformGcsPath())
                  .setFunctionName(options.getJavascriptTextTransformFunctionName())
                  .build());
    }

    /*
     * Step #4: Write converted records to Elasticsearch
     */
    udfOut.apply(
        "WriteToElasticsearch",
        WriteToElasticsearch.newBuilder()
            .setUserAgent("dataflow-bigquery-to-elasticsearch-template/v2")
            .setOptions(options.as(BigQueryToElasticsearchOptions.class))
            .build());

    return pipeline.run();
  }
}

What's next

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