Create a Bigtable external table

This page describes how to create a BigQuery permanent external table that can be used to query data stored in Bigtable. Querying data in Bigtable is available in all Bigtable locations.

Before you begin

Before you create an external table, gather some information and make sure you have permission to create the table.

Required roles

To create an external table to use to query your Bigtable data, you must be a principal in the Bigtable Admin (roles/bigtable.admin) role for the instance that contains the source table.

You also need the bigquery.tables.create BigQuery Identity and Access Management (IAM) permission.

Each of the following predefined Identity and Access Management roles includes this permission:

• BigQuery Data Editor (roles/bigquery.dataEditor)
• BigQuery Data Owner (roles/bigquery.dataOwner)
• BigQuery Admin (roles/bigquery.admin)

If you are not a principal in any of these roles, ask your administrator to grant you access or to create the external table for you.

For more information on Identity and Access Management roles and permissions in BigQuery, see Predefined roles and permissions. To view information on Bigtable permissions, see Access control with Identity and Access Management.

Create or identify a dataset

Before you create an external table, you must create a dataset to contain the external table. You can also use an existing dataset.

Optional: Designate or create a cluster

If you plan to frequently query the same data that serves your production application, we recommend that you designate a cluster in your Bigtable instance to be used solely for BigQuery analysis. This isolates the traffic from the cluster or clusters that you use for your application's reads and writes. To learn more about replication and creating instances that have more than one cluster, see About replication.

Identify or create an app profile

Before you create an external table, decide which Bigtable app profile that BigQuery should use to read the data. We recommend that you use an app profile that you designate for use only with BigQuery.

If you have a cluster in your Bigtable instance that is dedicated to BigQuery access, configure the app profile to use single-cluster routing to that cluster.

To learn how Bigtable app profiles work, see About app profiles. To see how to create a new app profile, see Create and configure app profiles.

Retrieve the Bigtable URI

To create an external table for a Bigtable data source, you must provide the Bigtable URI. To retrieve the Bigtable URI, do the following:

1. Open the Bigtable page in the console.

   Go to Bigtable

2. Retrieve the following details about your Bigtable data source:

   • Your project ID
   • Your Bigtable instance ID
   • The ID of the Bigtable app profile that you plan to use
   • The name of your Bigtable table

3. Compose the Bigtable URI using the following format, where:

   • project_id is the project containing your Bigtable instance
   • instance_id is the Bigtable instance ID
   • (Optional) app_profile is the app profile ID that you want to use
   • table_name is the name of the table you're querying

   https://googleapis.com/bigtable/projects/project_id/instances/instance_id[/appProfiles/app_profile]/tables/table_name

Create permanent external tables

When you create a permanent external table in BigQuery that is linked to a Bigtable data source, there are two options for specifying the format of the external table:

• If you are using the API or the bq command-line tool, you create a table definition file that defines the schema and metadata for the external table.
• If you are using SQL, you use the uri option of the CREATE EXTERNAL TABLE statement to specify the Bigtable table to pull data from, and the bigtable_options option to specify the table schema.

The external table data is not stored in the BigQuery table. Because the table is permanent, you can use dataset-level access controls to share the table with others who also have access to the underlying Bigtable data source.

To create a permanent table, choose one of the following methods.

CREATE EXTERNAL TABLE mydataset.BigtableTable
OPTIONS (
  format = 'CLOUD_BIGTABLE',
  uris = ['https://googleapis.com/bigtable/projects/myproject/instances/myBigtableInstance/tables/table1'],
  bigtable_options =
    """
    {
      columnFamilies: [
        {
          "familyId": "familyId1",
          "type": "INTEGER",
          "encoding": "BINARY"
        }
      ],
      readRowkeyAsString: true
    }
    """
);

import com.google.cloud.bigquery.BigQuery;
import com.google.cloud.bigquery.BigQueryException;
import com.google.cloud.bigquery.BigQueryOptions;
import com.google.cloud.bigquery.BigtableColumn;
import com.google.cloud.bigquery.BigtableColumnFamily;
import com.google.cloud.bigquery.BigtableOptions;
import com.google.cloud.bigquery.ExternalTableDefinition;
import com.google.cloud.bigquery.QueryJobConfiguration;
import com.google.cloud.bigquery.TableId;
import com.google.cloud.bigquery.TableInfo;
import com.google.cloud.bigquery.TableResult;
import com.google.common.collect.ImmutableList;
import org.apache.commons.codec.binary.Base64;

// Sample to queries an external bigtable data source using a permanent table
public class QueryExternalBigtablePerm {

  public static void main(String[] args) {
    // TODO(developer): Replace these variables before running the sample.
    String projectId = "MY_PROJECT_ID";
    String bigtableInstanceId = "MY_INSTANCE_ID";
    String bigtableTableName = "MY_BIGTABLE_NAME";
    String bigqueryDatasetName = "MY_DATASET_NAME";
    String bigqueryTableName = "MY_TABLE_NAME";
    String sourceUri =
        String.format(
            "https://googleapis.com/bigtable/projects/%s/instances/%s/tables/%s",
            projectId, bigtableInstanceId, bigtableTableName);
    String query = String.format("SELECT * FROM %s ", bigqueryTableName);
    queryExternalBigtablePerm(bigqueryDatasetName, bigqueryTableName, sourceUri, query);
  }

  public static void queryExternalBigtablePerm(
      String datasetName, String tableName, String sourceUri, String query) {
    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();

      BigtableColumnFamily.Builder statsSummary = BigtableColumnFamily.newBuilder();

      // Configuring Columns
      BigtableColumn connectedCell =
          BigtableColumn.newBuilder()
              .setQualifierEncoded(Base64.encodeBase64String("connected_cell".getBytes()))
              .setFieldName("connected_cell")
              .setType("STRING")
              .setEncoding("TEXT")
              .build();
      BigtableColumn connectedWifi =
          BigtableColumn.newBuilder()
              .setQualifierEncoded(Base64.encodeBase64String("connected_wifi".getBytes()))
              .setFieldName("connected_wifi")
              .setType("STRING")
              .setEncoding("TEXT")
              .build();
      BigtableColumn osBuild =
          BigtableColumn.newBuilder()
              .setQualifierEncoded(Base64.encodeBase64String("os_build".getBytes()))
              .setFieldName("os_build")
              .setType("STRING")
              .setEncoding("TEXT")
              .build();

      // Configuring column family and columns
      statsSummary
          .setColumns(ImmutableList.of(connectedCell, connectedWifi, osBuild))
          .setFamilyID("stats_summary")
          .setOnlyReadLatest(true)
          .setEncoding("TEXT")
          .setType("STRING")
          .build();

      // Configuring BigtableOptions is optional.
      BigtableOptions options =
          BigtableOptions.newBuilder()
              .setIgnoreUnspecifiedColumnFamilies(true)
              .setReadRowkeyAsString(true)
              .setColumnFamilies(ImmutableList.of(statsSummary.build()))
              .build();

      TableId tableId = TableId.of(datasetName, tableName);
      // Create a permanent table linked to the Bigtable table
      ExternalTableDefinition externalTable =
          ExternalTableDefinition.newBuilder(sourceUri, options).build();
      bigquery.create(TableInfo.of(tableId, externalTable));

      // Example query
      TableResult results = bigquery.query(QueryJobConfiguration.of(query));

      results
          .iterateAll()
          .forEach(row -> row.forEach(val -> System.out.printf("%s,", val.toString())));

      System.out.println("Query on external permanent table performed successfully.");
    } catch (BigQueryException | InterruptedException e) {
      System.out.println("Query not performed \n" + e.toString());
    }
  }
}

Query external tables

For more information, see Query Bigtable data.

Generated schema

By default, BigQuery exposes the values in a column family as an array of columns and within that, an array of values written at different timestamps. This schema preserves the natural layout of data in Bigtable, but SQL queries can be challenging. It is possible to promote columns to subfields within the parent column family and to read only the latest value from each cell. This represents both of the arrays in the default schema as scalar values.

Example

You are storing user profiles for a fictional social network. One data model for this might be a profile column family with individual columns for gender, age and email:

rowkey | profile:gender| profile:age| profile:email
-------| --------------| -----------| -------------
alice  | female        | 30         | [email protected]

Using the default schema, a GoogleSQL query to count the number of male users over 30 is:

SELECT
  COUNT(1)
FROM
  `dataset.table`
OMIT
  RECORD IF NOT SOME(profile.column.name = "gender"
    AND profile.column.cell.value = "male")
  OR NOT SOME(profile.column.name = "age"
    AND INTEGER(profile.column.cell.value) > 30)

Querying the data is less challenging if gender and age are exposed as sub- fields. To expose them as sub-fields, list gender and age as named columns in the profile column family when defining the table. You can also instruct BigQuery to expose the latest values from this column family because typically, only the latest value (and possibly the only value) is of interest.

After exposing the columns as sub-fields, the GoogleSQL query to count the number of male users over 30 is:

SELECT
  COUNT(1)
FROM
  `dataset.table`
WHERE
  profile.gender.cell.value="male"
  AND profile.age.cell.value > 30

Notice how gender and age are referenced directly as fields. The JSON configuration for this setup is:

  "bigtableOptions": {
    "readRowkeyAsString": "true",
    "columnFamilies": [
      {
          "familyId": "profile",
          "onlyReadLatest": "true",
          "columns": [
              {
                  "qualifierString": "gender",
                  "type": "STRING"
              },
              {
                  "qualifierString": "age",
                  "type": "INTEGER"
              }
          ]
      }
    ]
  }

Value encoding

Bigtable stores data as raw bytes, independent to data encoding. However, byte values are of limited use in SQL query analysis. Bigtable provides two basic types of scalar decoding: text and HBase-binary.

The text format assumes that all values are stored as alphanumeric text strings. For example, an integer 768 will be stored as the string "768". The binary encoding assumes that HBase's Bytes.toBytes class of methods were used to encode the data and applies an appropriate decoding method.

Supported regions and zones

Querying data in Bigtable is available in all supported Bigtable zones. You can find the list of zones here. For multi-cluster instances, BigQuery routes traffic based on Bigtable app profile settings.

Limitations

For information about limitations that apply to external tables, see External table limitations.

Scopes for Compute Engine instances

When you create a Compute Engine instance, you can specify a list of scopes for the instance. The scopes control the instance's access to Google Cloud products, including Bigtable. Applications running on the VM use the service account to call Google Cloud APIs.

If you set up a Compute Engine instance to run as a service account, and that service account accesses an external table linked to a Bigtable data source, you must add the Bigtable read-only data access scope (https://www.googleapis.com/auth/bigtable.data.readonly) to the instance. For more information, see Creating a Compute Engine instance for Bigtable.

For information on applying scopes to a Compute Engine instance, see Changing the service account and access scopes for an instance. For more information on Compute Engine service accounts, see Service accounts.

What's next

• Learn more about Bigtable schema design.
• Review the introduction to external data sources.