googleclouddataproc / spark-bigquery-connector Edit

BigQuery data source for Apache Spark: Read data from BigQuery into DataFrames, write DataFrames into BigQuery tables.

Version Matrix

Apache Spark SQL connector for Google BigQuery (Beta)

The connector supports reading Google BigQuery tables into Spark's DataFrames, and writing DataFrames back into BigQuery. This is done by using the Spark SQL Data Source API to communicate with BigQuery.

Beta Disclaimer

The BigQuery Storage API and this connector are in Beta and are subject to change.

Changes may include, but are not limited to:

  • Type conversion
  • Partitioning
  • Parameters

Breaking changes will be restricted to major and minor versions.

BigQuery Storage API

The Storage API streams data in parallel directly from BigQuery via gRPC without using Google Cloud Storage as an intermediary.

It has a number of advantages over using the previous export-based read flow that should generally lead to better read performance:

Direct Streaming

It does not leave any temporary files in Google Cloud Storage. Rows are read directly from BigQuery servers using the Arrow or Avro wire formats.


The new API allows column and predicate filtering to only read the data you are interested in.

Column Filtering

Since BigQuery is backed by a columnar datastore, it can efficiently stream data without reading all columns.

Predicate Filtering

The Storage API supports arbitrary pushdown of predicate filters. Connector version 0.8.0-beta and above support pushdown of arbitrary filters to Bigquery.

There is a known issue in Spark that does not allow pushdown of filters on nested fields. For example - filters like = "Sunnyvale" will not get pushdown to Bigquery.

Dynamic Sharding

The API rebalances records between readers until they all complete. This means that all Map phases will finish nearly concurrently. See this blog article on how dynamic sharding is similarly used in Google Cloud Dataflow.

See Configuring Partitioning for more details.


Enable the BigQuery Storage API

Follow these instructions.

Create a Google Cloud Dataproc cluster (Optional)

If you do not have an Apache Spark environment you can create a Cloud Dataproc cluster with pre-configured auth. The following examples assume you are using Cloud Dataproc, but you can use spark-submit on any cluster.

Any Dataproc cluster using the API needs the 'bigquery' or 'cloud-platform' scopes. Dataproc clusters have the 'bigquery' scope by default, so most clusters in enabled projects should work by default e.g.

gcloud dataproc clusters create "$MY_CLUSTER"

Downloading and Using the Connector

The latest version of the connector is publicly available in the following links:

version Link
Scala 2.11 gs://spark-lib/bigquery/spark-bigquery-with-dependencies_2.11-0.23.1.jar (HTTP link)
Scala 2.12 gs://spark-lib/bigquery/spark-bigquery-with-dependencies_2.12-0.23.1.jar (HTTP link)
Spark 2.4 gs://spark-lib/bigquery/spark-2.4-bigquery-0.23.1-preview.jar(HTTP link)

Note: If you are using scala jars please use the jar relevant to your Spark installation. Starting from Spark 2.4 onwards there is an option to use the Java only jar.

The connector is also available from the Maven Central repository. It can be used using the --packages option or the spark.jars.packages configuration property. Use the following value

version Connector Artifact
Scala 2.11
Scala 2.12
Spark 2.4

If you want to keep up with the latest version of the connector the following links can be used. Notice that for production environments where the connector version should be pinned, one of the above links should be used.

version Link
Scala 2.11 gs://spark-lib/bigquery/spark-bigquery-latest_2.11.jar (HTTP link)
Scala 2.12 gs://spark-lib/bigquery/spark-bigquery-latest_2.12.jar (HTTP link)
Spark 2.4 gs://spark-lib/bigquery/spark-2.4-bigquery-latest.jar (HTTP link)

Hello World Example

You can run a simple PySpark wordcount against the API without compilation by running

Dataproc image 1.5 and above

gcloud dataproc jobs submit pyspark --cluster "$MY_CLUSTER" \
  --jars gs://spark-lib/bigquery/spark-bigquery-latest_2.12.jar \

Dataproc image 1.4 and below

gcloud dataproc jobs submit pyspark --cluster "$MY_CLUSTER" \
  --jars gs://spark-lib/bigquery/spark-bigquery-latest_2.11.jar \

Example Codelab


The connector uses the cross language Spark SQL Data Source API:

Reading data from a BigQuery table

df = \
  .format("bigquery") \

or the Scala only implicit API:

val df ="bigquery-public-data.samples.shakespeare")

For more information, see additional code samples in Python, Scala and Java.

Reading data from a BigQuery query

The connector allows you to run any Standard SQL SELECT query on BigQuery and fetch its results directly to a Spark Dataframe. This is easily done as described in the following code sample:


sql = """
  SELECT tag, COUNT(*) c
  FROM (
    SELECT SPLIT(tags, '|') tags
    FROM `bigquery-public-data.stackoverflow.posts_questions` a
    WHERE EXTRACT(YEAR FROM creation_date)>=2014
  ), UNNEST(tags) tag
  LIMIT 10
df ="bigquery").load(sql)

Which yields the result

|       tag|      c|
|    python|1352904|
|      java|1218220|
|   android| 913638|
|       php| 911806|
|        c#| 905331|
|      html| 769499|
|    jquery| 608071|
|       css| 510343|
|       c++| 458938|

A second option is to use the query option like this:

df ="bigquery").option("query", sql).load()

Notice that the execution should be faster as only the result is transmitted over the wire. In a similar fashion the queries can include JOINs more efficiently then running joins on Spark or use other BigQuery features such as subqueries, BigQuery user defined functions, wildcard tables, BigQuery ML and more.

In order to use this feature the following configurations MUST be set:

  • viewsEnabled must be set to true.
  • materializationDataset must be set to a dataset where the GCP user has table creation permission. materializationProject is optional.

Note: As mentioned in the BigQuery documentation, the queried tables must be in the same location as the materializationDataset. Also, if the tables in the SQL statement are from projects other than the parentProject then use the fully qualified table name i.e. [project].[dataset].[table].

Important: This feature is implemented by running the query on BigQuery and saving the result into a temporary table, of which Spark will read the results from. This may add additional costs on your BigQuery account.

Reading From Views

The connector has a preliminary support for reading from BigQuery views. Please note there are a few caveats:

  • BigQuery views are not materialized by default, which means that the connector needs to materialize them before it can read them. This process affects the read performance, even before running any collect() or count() action.
  • The materialization process can also incur additional costs to your BigQuery bill.
  • By default, the materialized views are created in the same project and dataset. Those can be configured by the optional materializationProject and materializationDataset options, respectively. These options can also be globally set by calling spark.conf.set(...) before reading the views.
  • Reading from views is disabled by default. In order to enable it, either set the viewsEnabled option when reading the specific view (.option("viewsEnabled", "true")) or set it globally by calling spark.conf.set("viewsEnabled", "true").
  • As mentioned in the BigQuery documentation, the materializationDataset should be in same location as the view.

Writing data to BigQuery

Direct write using the BigQuery Storage Write API

The Spark 2.4 dedicated connector supports writing directly to BigQuery without first writing to GCS, using the BigQuery Storage Write API to write data directly to BigQuery. In order to enable this option, please set the writeMethod option to direct, as shown below:

df.write \
  .format("bigquery") \
  .option("writeMethod", "direct") \

Important: Please refer to the data ingestion pricing page regarding the BigQuery Storage Write API pricing.

Indirect write

This method is supported by all the connector. In this method the data is written first to GCS and then it is loaded it to BigQuery. A GCS bucket must be configured to indicate the temporary data location.

df.write \
  .format("bigquery") \
  .option("temporaryGcsBucket","some-bucket") \

The data is temporarily stored using the Apache Parquet, Apache ORC or Apache Avro formats.

The GCS bucket and the format can also be set globally using Spark's RuntimeConfig like this:

df.write \
  .format("bigquery") \

When streaming a DataFrame to BigQuery, each batch is written in the same manner as a non-streaming DataFrame. Note that a HDFS compatible checkpoint location (eg: path/to/HDFS/dir or gs://checkpoint-bucket/checkpointDir) must be specified.

df.writeStream \
  .format("bigquery") \
  .option("temporaryGcsBucket","some-bucket") \
  .option("checkpointLocation", "some-location") \
  .option("table", "dataset.table")

Important: The connector does not configure the GCS connector, in order to avoid conflict with another GCS connector, if exists. In order to use the write capabilities of the connector, please configure the GCS connector on your cluster as explained here.


The API Supports a number of options to configure the read

Property Meaning Usage
table The BigQuery table in the format [[project:]dataset.]table. It is recommended to use the path parameter of load()/save() instead. This option has been deprecated and will be removed in a future version.
dataset The dataset containing the table. This option should be used with standard table and views, but not when loading query results.
(Optional unless omitted in table)
project The Google Cloud Project ID of the table. This option should be used with standard table and views, but not when loading query results.
(Optional. Defaults to the project of the Service Account being used)
parentProject The Google Cloud Project ID of the table to bill for the export.
(Optional. Defaults to the project of the Service Account being used)
maxParallelism The maximal number of partitions to split the data into. Actual number may be less if BigQuery deems the data small enough. If there are not enough executors to schedule a reader per partition, some partitions may be empty.
Important: The old parameter (parallelism) is still supported but in deprecated mode. It will ve removed in version 1.0 of the connector.
(Optional. Defaults to one partition per 400MB. See Configuring Partitioning.)
viewsEnabled Enables the connector to read from views and not only tables. Please read the relevant section before activating this option.
(Optional. Defaults to false)
materializationProject The project id where the materialized view is going to be created
(Optional. Defaults to view's project id)
materializationDataset The dataset where the materialized view is going to be created. This dataset should be in same location as the view or the queried tables.
(Optional. Defaults to view's dataset)
materializationExpirationTimeInMinutes The expiration time of the temporary table holding the materialized data of a view or a query, in minutes. Notice that the connector may re-use the temporary table due to the use of local cache and in order to reduce BigQuery computation, so very low values may cause errors. The value must be a positive integer.
(Optional. Defaults to 1440, or 24 hours)
readDataFormat Data Format for reading from BigQuery. Options : ARROW, AVRO Unsupported Arrow filters are not pushed down and results are filtered later by Spark. (Currently Arrow does not suport disjunction across columns).
(Optional. Defaults to ARROW)
optimizedEmptyProjection The connector uses an optimized empty projection (select without any columns) logic, used for count() execution. This logic takes the data directly from the table metadata or performs a much efficient `SELECT COUNT(*) WHERE...` in case there is a filter. You can cancel the use of this logic by setting this option to false.
(Optional, defaults to true)
pushAllFilters If set to true, the connector pushes all the filters Spark can delegate to BigQuery Storage API. This reduces amount of data that needs to be sent from BigQuery Storage API servers to Spark clients.
(Optional, defaults to true)
createDisposition Specifies whether the job is allowed to create new tables. The permitted values are:
  • CREATE_IF_NEEDED - Configures the job to create the table if it does not exist.
  • CREATE_NEVER - Configures the job to fail if the table does not exist.
This option takes place only in case Spark has decided to write data to the table based on the SaveMode.
(Optional. Default to CREATE_IF_NEEDED).
writeMethod Used only by the Spark 2.4 dedicated connector. Controls the method in which the data is written to BigQuery. Available values are direct to use the BigQuery Storage Write API and indirect which writes the data first to GCS and then triggers a BigQuery load operation. See more here
(Optional, defaults to indirect)
Write (supported only by the Spark 2.4 dedicated connector)
temporaryGcsBucket The GCS bucket that temporarily holds the data before it is loaded to BigQuery. Required unless set in the Spark configuration (spark.conf.set(...)). Write
persistentGcsBucket The GCS bucket that holds the data before it is loaded to BigQuery. If informed, the data won't be deleted after write data into BigQuery. Write
persistentGcsPath The GCS path that holds the data before it is loaded to BigQuery. Used only with persistentGcsBucket. Write
intermediateFormat The format of the data before it is loaded to BigQuery, values can be either "parquet","orc" or "avro". In order to use the Avro format, the spark-avro package must be added in runtime.
(Optional. Defaults to parquet). On write only.
datePartition The date partition the data is going to be written to. Should be given in the format YYYYMMDD. Can be used to overwrite the data of a single partition, like this:
  .option("datePartition", "YYYYMMDD")

(Optional). On write only.
partitionField If field is specified together with `partitionType`, the table is partitioned by this field. The field must be a top-level TIMESTAMP or DATE field. Its mode must be NULLABLE or REQUIRED. If the option is not set for a partitioned table, then the table will be partitioned by pseudo column, referenced via either'_PARTITIONTIME' as TIMESTAMP type, or '_PARTITIONDATE' as DATE type.
partitionExpirationMs Number of milliseconds for which to keep the storage for partitions in the table. The storage in a partition will have an expiration time of its partition time plus this value.
partitionType The only type supported is DAY, which will generate one partition per day. This option is mandatory for a target table to be partitioned.
(Optional. Defaults to DAY if PartitionField is specified).
clusteredFields Comma separated list of non-repeated, top level columns. Clustering is only supported for partitioned tables
allowFieldAddition Adds the ALLOW_FIELD_ADDITION SchemaUpdateOption to the BigQuery LoadJob. Allowed values are true and false.
(Optional. Default to false).
allowFieldRelaxation Adds the ALLOW_FIELD_RELAXATION SchemaUpdateOption to the BigQuery LoadJob. Allowed values are true and false.
(Optional. Default to false).
proxyAddress Address of the proxy server. The proxy must be a HTTP proxy and address should be in the `host:port` format. Can be alternatively set in the Spark configuration (spark.conf.set(...)) or in Hadoop Configuration (
(Optional. Required only if connecting to GCP via proxy.)
proxyUsername The userName used to connect to the proxy. Can be alternatively set in the Spark configuration (spark.conf.set(...)) or in Hadoop Configuration (
(Optional. Required only if connecting to GCP via proxy with authentication.)
proxyPassword The password used to connect to the proxy. Can be alternatively set in the Spark configuration (spark.conf.set(...)) or in Hadoop Configuration (
(Optional. Required only if connecting to GCP via proxy with authentication.)
httpMaxRetry The maximum number of retries for the low-level HTTP requests to BigQuery. Can be alternatively set in the Spark configuration (spark.conf.set("httpMaxRetry", ...)) or in Hadoop Configuration (
(Optional. Default is 10)
httpConnectTimeout The timeout in milliseconds to establish a connection with BigQuery. Can be alternatively set in the Spark configuration (spark.conf.set("httpConnectTimeout", ...)) or in Hadoop Configuration (
(Optional. Default is 60000 ms. 0 for an infinite timeout, a negative number for 20000)
httpReadTimeout The timeout in milliseconds to read data from an established connection. Can be alternatively set in the Spark configuration (spark.conf.set("httpReadTimeout", ...)) or in Hadoop Configuration (
(Optional. Default is 60000 ms. 0 for an infinite timeout, a negative number for 20000)
arrowCompressionCodec Compression codec while reading from a BigQuery table when using Arrow format. Options : ZSTD (Zstandard compression), LZ4_FRAME (, COMPRESSION_UNSPECIFIED. The recommended compression codec is ZSTD while using Java.
(Optional. Defaults to COMPRESSION_UNSPECIFIED which means no compression will be used)

Options can also be set outside of the code, using the --conf parameter of spark-submit or --properties parameter of the gcloud dataproc submit spark. In order to use this, prepend the prefix spark.datasource.bigquery. to any of the options, for example spark.conf.set("temporaryGcsBucket", "some-bucket") can also be set as --conf spark.datasource.bigquery.temporaryGcsBucket=some-bucket.

Data types

With the exception of DATETIME and TIME all BigQuery data types directed map into the corresponding Spark SQL data type. Here are all of the mappings:

BigQuery Standard SQL Data Type Spark SQL

Data Type

BOOL BooleanType
INT64 LongType
FLOAT64 DoubleType
NUMERIC DecimalType This preserves NUMERIC's full 38 digits of precision and 9 digits of scope.
BIGNUMERIC BigNumericUDT (UserDefinedType) Scala/Java: BigNumericUDT DataType internally uses java.math.BigDecimal to hold the BigNumeric data.

Python: BigNumericUDT DataType internally used python's Decimal class to hold the BigNumeric data.

STRING StringType
BYTES BinaryType
STRUCT StructType
ARRAY ArrayType
TIMESTAMP TimestampType
DATE DateType
DATETIME StringType Spark has no DATETIME type. Casting to TIMESTAMP uses a configured TimeZone, which defaults to the local timezone (UTC in GCE / Dataproc).

We are considering adding an optional TimeZone property to allow automatically converting to TimeStamp, this would be consistent with Spark's handling of CSV/JSON (except they always try to convert when inferring schema, and default to the local timezone)

TIME LongType Spark has no TIME type. The generated longs, which indicate microseconds since midnight can be safely cast to TimestampType, but this causes the date to be inferred as the current day. Thus times are left as longs and user can cast if they like.

When casting to Timestamp TIME have the same TimeZone issues as DATETIME

Spark ML Data Types Support

The Spark ML Vector and Matrix are supported, including their dense and sparse versions. The data is saved as a BigQuery RECORD. Notice that a suffix is added to the field's description which includes the spark type of the field.

In order to write those types to BigQuery, use the ORC or Avro intermediate format, and have them as column of the Row (i.e. not a field in a struct).

BigNumeric support

BigQuery's BigNumeric has a precision of 76.76 (the 77th digit is partial) and scale of 38. Since this precision and scale is beyond spark's DecimalType (38 scale and 38 precision) support, the BigNumeric DataType is converted into spark's UserDefinedType. The BigNumeric data can be accessed via BigNumericUDT DataType which internally uses java.math.BigDecimal to hold the BigNumeric data. The data can be read in either AVRO or ARROW formats.

In order to write BigNumericUDT to BigQuery, use either ORC or PARQUET intermediate formats (currently we do not support AVRO). Notice that the data gets written to BigQuery as String.

Code examples:


import org.apache.spark.bigquery.BigNumeric

val df =

val rows: Array[java.math.BigDecimal] = df
  .map(row => row.get("BIG_NUMERIC_COLUMN").asInstanceOf[BigNumeric].getNumber)

rows.foreach(value => System.out.println("BigNumeric value  " + value.toPlainString))

Python: Spark's UserDefinedType needs a separate implementation for Python. Corresponding python class(s) should be provided as config params while creating the job or added during runtime. See examples below:

  1. Adding python files while launching pyspark
# use appropriate version for jar depending on the scala version
pyspark --jars gs://spark-lib/bigquery/spark-bigquery-with-dependencies_2.11-0.23.1.jar
  --py-files gs://spark-lib/bigquery/
  --files gs://spark-lib/bigquery/
  1. Adding python files in Jupyter Notebook
from pyspark.sql import SparkSession
# use appropriate version for jar depending on the scala version
spark = SparkSession.builder\
  .config('spark.jars', 'gs://spark-lib/bigquery/spark-bigquery-with-dependencies_2.11-0.23.1.jar')\
  .config('spark.submit.pyFiles', 'gs://spark-lib/bigquery/')\
  .config('spark.files', 'gs://spark-lib/bigquery/')\
  1. Adding Python files during runtime
# use appropriate version for jar depending on the scala version
spark = SparkSession.builder\
  .config('spark.jars', 'gs://spark-lib/bigquery/spark-bigquery-with-dependencies_2.11-0.23.1.jar')\


Usage Example:

df ="bigquery").load({project}.{dataset}.{table_name})
data ={big_numeric_column_name}).collect()

for row in data:
  bigNumeric = row[{big_numeric_column_name}]
  # bigNumeric.number is instance of python's Decimal class

In case the above code throws ModuleNotFoundError, please add the following code before reading the BigNumeric data.

    import pkg_resources

except ImportError:
    import pkgutil

    __path__ = pkgutil.extend_path(__path__, __name__)


The connector automatically computes column and pushdown filters the DataFrame's SELECT statement e.g."bigquery-public-data:samples.shakespeare")
  .where("word = 'Hamlet' or word = 'Claudius'")

filters to the column word and pushed down the predicate filter word = 'hamlet' or word = 'Claudius'.

If you do not wish to make multiple read requests to BigQuery, you can cache the DataFrame before filtering e.g.:

val cachedDF ="bigquery-public-data:samples.shakespeare").cache()
val rows ="word")
  .where("word = 'Hamlet'")
// All of the table was cached and this doesn't require an API call
val otherRows ="word_count")
  .where("word = 'Romeo'")

You can also manually specify the filter option, which will override automatic pushdown and Spark will do the rest of the filtering in the client.

Partitioned Tables

The pseudo columns _PARTITIONDATE and _PARTITIONTIME are not part of the table schema. Therefore in order to query by the partitions of partitioned tables do not use the where() method shown above. Instead, add a filter option in the following manner:

val df ="bigquery")
  .option("filter", "_PARTITIONDATE > '2019-01-01'")

Configuring Partitioning

By default the connector creates one partition per 400MB in the table being read (before filtering). This should roughly correspond to the maximum number of readers supported by the BigQuery Storage API. This can be configured explicitly with the maxParallelism property. BigQuery may limit the number of partitions based on server constraints.

Using in Jupyter Notebooks

The connector can be used in Jupyter notebooks even if it is not installed on the Spark cluster. It can be added as an external jar in using the following code:


from pyspark.sql import SparkSession
spark = SparkSession.builder\
  .config("spark.jars.packages", "")\
df ="bigquery")\


val spark = SparkSession.builder
  .config("spark.jars.packages", "")
val df ="bigquery")

In case Spark cluster is using Scala 2.12 (it's optional for Spark 2.4.x, mandatory in 3.0.x), then the relevant package is In order to know which Scala version is used, please run the following code:





Compiling against the connector

Unless you wish to use the implicit Scala API"TABLE_ID"), there is no need to compile against the connector.

To include the connector in your project:




libraryDependencies += "" %% "spark-bigquery-with-dependencies" % "0.23.1"


What is the Pricing for the Storage API?

See the BigQuery pricing documentation.

I have very few partitions

You can manually set the number of partitions with the maxParallelism property. BigQuery may provide fewer partitions than you ask for. See Configuring Partitioning.

You can also always repartition after reading in Spark.

How do I authenticate outside GCE / Dataproc?

Use a service account JSON key and GOOGLE_APPLICATION_CREDENTIALS as described here.

Credentials can also be provided explicitly either as a parameter or from Spark runtime configuration. It can be passed in as a base64-encoded string directly, or a file path that contains the credentials (but not both).

Example:"bigquery").option("credentials", "<SERVICE_ACCOUNT_JSON_IN_BASE64>")


spark.conf.set("credentials", "<SERVICE_ACCOUNT_JSON_IN_BASE64>")

Alternatively, specify the credentials file name."bigquery").option("credentialsFile", "</path/to/key/file>")


spark.conf.set("credentialsFile", "</path/to/key/file>")

Another alternative to passing the credentials, is to pass the access token used for authenticating the API calls to the Google Cloud Platform APIs. You can get the access token by running gcloud auth application-default print-access-token."bigquery").option("gcpAccessToken", "<acccess-token>")


spark.conf.set("gcpAccessToken", "<access-token>")

How do I connect to GCP/BigQuery via Proxy?

To connect to a forward proxy and to authenticate the user credentials, configure the following options.

proxyAddress: Address of the proxy server. The proxy must be an HTTP proxy and address should be in the host:port format.

proxyUsername: The userName used to connect to the proxy.

proxyPassword: The password used to connect to the proxy.

val df ="bigquery")
  .option("proxyAddress", "http://my-proxy:1234")
  .option("proxyUsername", "my-username")
  .option("proxyPassword", "my-password")

The same proxy parameters can also be set globally using Spark's RuntimeConfig like this:

spark.conf.set("proxyAddress", "http://my-proxy:1234")
spark.conf.set("proxyUsername", "my-username")
spark.conf.set("proxyPassword", "my-password")

val df ="bigquery")

You can set the following in the hadoop configuration as well. to "proxyAddress"), to "proxyUsername") and to "proxyPassword").

If the same parameter is set at multiple places the order of priority is as follows:

option("key", "value") > spark.conf > hadoop configuration