BigQuery

BigQuery is a completely serverless and cost-effective enterprise data warehouse that works across clouds and scales with your data, with BI, machine learning and AI built in.

The BigQuery Wrapper allows you to read and write data from BigQuery within your Postgres database.

Preparation#

Before you can query BigQuery, you need to enable the Wrappers extension and store your credentials in Postgres.

Enable Wrappers#

Make sure the wrappers extension is installed on your database:

1
create extension if not exists wrappers with schema extensions;

Enable the BigQuery Wrapper#

Enable the bigquery_wrapper FDW:

1
create foreign data wrapper bigquery_wrapper
2
  handler big_query_fdw_handler
3
  validator big_query_fdw_validator;

Store your credentials (optional)#

By default, Postgres stores FDW credentials inside pg_catalog.pg_foreign_server in plain text. Anyone with access to this table will be able to view these credentials. Wrappers is designed to work with Vault, which provides an additional level of security for storing credentials. We recommend using Vault to store your credentials.

1
-- Save your BigQuery service account json in Vault and retrieve the created `key_id`
2
select vault.create_secret(
3
  '
4
    {
5
      "type": "service_account",
6
      "project_id": "your_gcp_project_id",
7
      "private_key_id": "your_private_key_id",
8
      "private_key": "-----BEGIN PRIVATE KEY-----\n...\n-----END PRIVATE KEY-----\n",
9
      ...
10
    }
11
  ',
12
  'bigquery',
13
  'BigQuery service account json for Wrappers'
14
);

Connecting to BigQuery#

We need to provide Postgres with the credentials to connect to BigQuery, and any additional options. We can do this using the create server command:

1
create server bigquery_server
2
  foreign data wrapper bigquery_wrapper
3
  options (
4
    sa_key_id '<key_ID>', -- The Key ID from above.
5
    project_id 'your_gcp_project_id',
6
    dataset_id 'your_gcp_dataset_id'
7
  );

Create a schema#

We recommend creating a schema to hold all the foreign tables:

1
create schema if not exists bigquery;

Options#

The following options are available when creating BigQuery foreign tables:

• table - Source table or view name in BigQuery, required
• location - Source table location (default: 'US')
• timeout - Query request timeout in milliseconds (default: 30000)
• rowid_column - Primary key column name (required for data modification)

You can also use a subquery as the table option:

1
table '(select * except(props), to_json_string(props) as props from `my_project.my_dataset.my_table`)'

Note: When using subquery, full qualified table name must be used.

Entites#

Tables#

The BigQuery Wrapper supports data reads and writes from BigQuery tables and views.

Operations#

Usage#

1
create foreign table bigquery.my_bigquery_table (
2
  id bigint,
3
  name text,
4
  ts timestamp
5
)
6
  server bigquery_server
7
  options (
8
    table 'people',
9
    location 'EU'
10
  );

Notes#

• Supports where, order by, limit and aggregate clause pushdown
• When using rowid_column, it must be specified for data modification operations
• Data in the streaming buffer cannot be updated or deleted until the buffer is flushed (up to 90 minutes)

Query Pushdown Support#

This FDW supports where, order by and limit clause pushdown.

Aggregate Pushdown#

The FDW pushes common aggregate queries down to BigQuery so the aggregation runs remotely and only the final result rows are transferred to Postgres. This is much faster than fetching every row and aggregating locally, especially over large tables — and on BigQuery it also reduces the bytes scanned billed to your project.

Supported aggregates — count(*), count(col), count(distinct col), sum(col), avg(col), min(col), max(col).

Supported shapes — scalar aggregates, group by over plain columns, with or without a where clause. Pushdown also works when the foreign table option is a sub-query.

1
-- All of these run as a single aggregate query on BigQuery:
2
select count(*) from bigquery.my_table;
3
select id, sum(amount) from bigquery.my_table group by id;
4
select count(distinct name) from bigquery.my_table where id = 1;

Cases that are not pushed down — the query still returns the correct result, but the aggregation happens in Postgres after fetching the rows:

• The query has a having clause
• The aggregate has a filter (where …) clause
• A distinct modifier is used on anything other than count
• The aggregate's argument is not a plain column (for example sum(a + 1))
• A group by item is not a plain column (for example group by id + 1)
• The aggregate function is not in the list above (for example stddev, string_agg)

Inserting Rows & the Streaming Buffer#

This foreign data wrapper uses BigQuery’s insertAll API method to create a streamingBuffer with an associated partition time. Within that partition time, the data cannot be updated, deleted, or fully exported. Only after the time has elapsed (up to 90 minutes according to BigQuery’s documentation), can you perform operations.

If you attempt an UPDATE or DELETE statement on rows while in the streamingBuffer, you will get an error of UPDATE or DELETE statement over table datasetName - note that tableName would affect rows in the streaming buffer, which is not supported.

Supported Data Types#

Limitations#

This section describes important limitations and considerations when using this FDW:

• Large result sets may experience network latency during data transfer
• Data in streaming buffer cannot be modified for up to 90 minutes
• Only supports specific data type mappings between Postgres and BigQuery
• Materialized views using foreign tables may fail during logical backups

Examples#

Some examples on how to use BigQuery foreign tables.

Let's prepare the source table in BigQuery first:

1
-- Run below SQLs on BigQuery to create source table
2
create table your_project_id.your_dataset_id.people (
3
  id int64,
4
  name string,
5
  ts timestamp,
6
  props jsonb
7
);
8
9
-- Add some test data
10
insert into your_project_id.your_dataset_id.people values
11
  (1, 'Luke Skywalker', current_timestamp(), parse_json('{"coordinates":[10,20],"id":1}')),
12
  (2, 'Leia Organa', current_timestamp(), null),
13
  (3, 'Han Solo', current_timestamp(), null);

Basic example#

This example will create a "foreign table" inside your Postgres database called people and query its data:

1
create foreign table bigquery.people (
2
  id bigint,
3
  name text,
4
  ts timestamp,
5
  props jsonb
6
)
7
  server bigquery_server
8
  options (
9
    table 'people',
10
    location 'EU'
11
  );
12
13
select * from bigquery.people;

Data modify example#

This example will modify data in a "foreign table" inside your Postgres database called people, note that rowid_column option is mandatory:

1
create foreign table bigquery.people (
2
  id bigint,
3
  name text,
4
  ts timestamp,
5
  props jsonb
6
)
7
  server bigquery_server
8
  options (
9
    table 'people',
10
    location 'EU',
11
    rowid_column 'id'
12
  );
13
14
-- insert new data
15
insert into bigquery.people(id, name, ts, props)
16
values (4, 'Yoda', '2023-01-01 12:34:56', '{"coordinates":[10,20],"id":1}'::jsonb);
17
18
-- update existing data
19
update bigquery.people
20
set name = 'Anakin Skywalker', props = '{"coordinates":[30,40],"id":42}'::jsonb
21
where id = 1;
22
23
-- delete data
24
delete from bigquery.people
25
where id = 2;

Aggregate Query Examples#

These examples assume an orders table on BigQuery and a matching foreign table on Postgres:

1
-- Run on BigQuery
2
create table your_project_id.your_dataset_id.orders (
3
  id       int64,
4
  user_id  int64,
5
  amount   numeric,
6
  status   string
7
);
8
9
insert into your_project_id.your_dataset_id.orders values
10
  (1, 1, 100.0, 'paid'),
11
  (2, 1,  50.0, 'paid'),
12
  (3, 2, 200.0, 'pending'),
13
  (4, 2,  75.0, 'paid'),
14
  (5, 3, 300.0, 'paid');

1
-- Foreign table on Postgres
2
create foreign table bigquery.orders (
3
  id      bigint,
4
  user_id bigint,
5
  amount  numeric,
6
  status  text
7
)
8
  server bigquery_server
9
  options (
10
    table 'orders'
11
  );

Each query below runs a single aggregate query against BigQuery and returns just the result rows:

1
-- Total order count
2
select count(*) from bigquery.orders;
3
4
-- Total revenue from paid orders
5
select sum(amount) from bigquery.orders where status = 'paid';
6
7
-- Per-user order count and revenue
8
select user_id, count(*) as orders, sum(amount) as revenue
9
from bigquery.orders
10
group by user_id
11
order by user_id;
12
13
-- Smallest and largest order
14
select min(amount), max(amount) from bigquery.orders;
15
16
-- Number of distinct users who placed an order
17
select count(distinct user_id) from bigquery.orders;
18
19
-- Average order value per status
20
select status, avg(amount) as avg_amount
21
from bigquery.orders
22
group by status;