NebulaGraph Spark Connector¶

NebulaGraph Spark Connector is a Spark connector application for reading and writing NebulaGraph data in Spark standard format. NebulaGraph Spark Connector consists of two parts: Reader and Writer.

Reader

Provides a Spark SQL interface. This interface can be used to read NebulaGraph data. It reads one vertex or edge type data at a time and assemble the result into a Spark DataFrame.

Writer

Provides a Spark SQL interface. This interface can be used to write DataFrames into NebulaGraph in a row-by-row or batch-import way.

For more information, see NebulaGraph Spark Connector.

Version compatibility¶

The correspondence between the NebulaGraph Spark Connector version, the NebulaGraph core version and the Spark version is as follows.

Spark Connector version	NebulaGraph version	Spark version
nebula-spark-connector_3.0-3.0-SNAPSHOT.jar	nightly	3.x
nebula-spark-connector_2.2-3.0-SNAPSHOT.jar	nightly	2.2.x
nebula-spark-connector-3.0-SNAPSHOT.jar	nightly	2.4.x
nebula-spark-connector_2.2-3.4.0.jar	3.x	2.2.x
nebula-spark-connector-3.4.0.jar	3.x	2.4.x
nebula-spark-connector_2.2-3.3.0.jar	3.x	2.2.x
nebula-spark-connector-3.3.0.jar	3.x	2.4.x
nebula-spark-connector-3.0.0.jar	3.x	2.4.x
nebula-spark-connector-2.6.1.jar	2.6.0, 2.6.1	2.4.x
nebula-spark-connector-2.6.0.jar	2.6.0, 2.6.1	2.4.x
nebula-spark-connector-2.5.1.jar	2.5.0, 2.5.1	2.4.x
nebula-spark-connector-2.5.0.jar	2.5.0, 2.5.1	2.4.x
nebula-spark-connector-2.1.0.jar	2.0.0, 2.0.1	2.4.x
nebula-spark-connector-2.0.1.jar	2.0.0, 2.0.1	2.4.x
nebula-spark-connector-2.0.0.jar	2.0.0, 2.0.1	2.4.x

Use cases¶

NebulaGraph Spark Connector applies to the following scenarios:

Migrate data between different NebulaGraph clusters.

Migrate data between different graph spaces in the same NebulaGraph cluster.

Migrate data between NebulaGraph and other data sources.

Graph computing with NebulaGraph Algorithm.

Benefits¶

The features of NebulaGraph Spark Connector 3.4.0 are as follows:

Supports multiple connection settings, such as timeout period, number of connection retries, number of execution retries, etc.

Supports multiple settings for data writing, such as setting the corresponding column as vertex ID, starting vertex ID, destination vertex ID or attributes.

Supports non-attribute reading and full attribute reading.

Supports reading NebulaGraph data into VertexRDD and EdgeRDD, and supports non-Long vertex IDs.

Unifies the extended data source of SparkSQL, and uses DataSourceV2 to extend NebulaGraph data.

Three write modes, insert, update and delete, are supported. insert mode will insert (overwrite) data, update mode will only update existing data, and delete mode will only delete data.

Release note¶

Release

Get NebulaGraph Spark Connector¶

Compile package¶

Clone repository nebula-spark-connector.

$ git clone -b release-3.4 https://github.com/vesoft-inc/nebula-spark-connector.git

2. Enter the nebula-spark-connector directory.

Compile package. The procedure varies with Spark versions.

Note

Spark of the corresponding version has been installed.

- Spark 2.4

```bash
$ mvn clean package -Dmaven.test.skip=true -Dgpg.skip -Dmaven.javadoc.skip=true -pl nebula-spark-connector -am -Pscala-2.11 -Pspark-2.4
```

- Spark 2.2

```bash
$ mvn clean package -Dmaven.test.skip=true -Dgpg.skip -Dmaven.javadoc.skip=true -pl nebula-spark-connector_2.2 -am -Pscala-2.11 -Pspark-2.2
```

- Spark 3.x

```bash
$ mvn clean package -Dmaven.test.skip=true -Dgpg.skip -Dmaven.javadoc.skip=true -pl nebula-spark-connector_3.0 -am -Pscala-2.12 -Pspark-3.0
```

After compilation, a similar file nebula-spark-connector-3.4.0-SHANPSHOT.jar is generated in the directory target of the folder.

Download maven remote repository¶

Download

How to use¶

When using NebulaGraph Spark Connector to reading and writing NebulaGraph data, You can refer to the following code.

# Read vertex and edge data from NebulaGraph.
spark.read.nebula().loadVerticesToDF()
spark.read.nebula().loadEdgesToDF()

# Write dataframe data into NebulaGraph as vertex and edges.
dataframe.write.nebula().writeVertices()
dataframe.write.nebula().writeEdges()

nebula() receives two configuration parameters, including connection configuration and read-write configuration.

Reading data from NebulaGraph¶

val config = NebulaConnectionConfig
  .builder()
  .withMetaAddress("127.0.0.1:9559")
  .withConenctionRetry(2)
  .withExecuteRetry(2)
  .withTimeout(6000)
  .build()

val nebulaReadVertexConfig: ReadNebulaConfig = ReadNebulaConfig
  .builder()
  .withSpace("test")
  .withLabel("person")
  .withNoColumn(false)
  .withReturnCols(List("birthday"))
  .withLimit(10)
  .withPartitionNum(10)
  .build()
val vertex = spark.read.nebula(config, nebulaReadVertexConfig).loadVerticesToDF()

val nebulaReadEdgeConfig: ReadNebulaConfig = ReadNebulaConfig
  .builder()
  .withSpace("test")
  .withLabel("knows")
  .withNoColumn(false)
  .withReturnCols(List("degree"))
  .withLimit(10)
  .withPartitionNum(10)
  .build()
val edge = spark.read.nebula(config, nebulaReadEdgeConfig).loadEdgesToDF()

NebulaConnectionConfig is the configuration for connecting to the nebula graph, as described below.

Parameter	Required	Description
`withMetaAddress`	Yes	Specifies the IP addresses and ports of all Meta Services. Separate multiple addresses with commas. The format is `ip1:port1,ip2:port2,...`. Read data is no need to configure `withGraphAddress`.
`withConnectionRetry`	No	The number of retries that the NebulaGraph Java Client connected to the NebulaGraph. The default value is `1`.
`withExecuteRetry`	No	The number of retries that the NebulaGraph Java Client executed query statements. The default value is `1`.
`withTimeout`	No	The timeout for the NebulaGraph Java Client request response. The default value is `6000`, Unit: ms.

ReadNebulaConfig is the configuration to read NebulaGraph data, as described below.

Parameter	Required	Description
`withSpace`	Yes	NebulaGraph space name.
`withLabel`	Yes	The Tag or Edge type name within the NebulaGraph space.
`withNoColumn`	No	Whether the property is not read. The default value is `false`, read property. If the value is `true`, the property is not read, the `withReturnCols` configuration is invalid.
`withReturnCols`	No	Configures the set of properties for vertex or edges to read. the format is `List(property1,property2,...)`, The default value is `List()`, indicating that all properties are read.
`withLimit`	No	Configure the number of rows of data read from the server by the NebulaGraph Java Storage Client at a time. The default value is `1000`.
`withPartitionNum`	No	Configures the number of Spark partitions to read the NebulaGraph data. The default value is `100`. This value should not exceed the number of slices in the graph space (partition_num).

Write data into NebulaGraph¶

Note

The values of columns in a dataframe are automatically written to the NebulaGraph as property values.

val config = NebulaConnectionConfig
  .builder()
  .withMetaAddress("127.0.0.1:9559")
  .withGraphAddress("127.0.0.1:9669")
  .withConenctionRetry(2)
  .build()

val nebulaWriteVertexConfig: WriteNebulaVertexConfig = WriteNebulaVertexConfig      
  .builder()
  .withSpace("test")
  .withTag("person")
  .withVidField("id")
  .withVidPolicy("hash")
  .withVidAsProp(true)
  .withUser("root")
  .withPasswd("nebula")
  .withBatch(1000)
  .build()    
df.write.nebula(config, nebulaWriteVertexConfig).writeVertices()

val nebulaWriteEdgeConfig: WriteNebulaEdgeConfig = WriteNebulaEdgeConfig      
  .builder()
  .withSpace("test")
  .withEdge("friend")
  .withSrcIdField("src")
  .withSrcPolicy(null)
  .withDstIdField("dst")
  .withDstPolicy(null)
  .withRankField("degree")
  .withSrcAsProperty(true)
  .withDstAsProperty(true)
  .withRankAsProperty(true)
  .withUser("root")
  .withPasswd("nebula")
  .withBatch(1000)
  .build()
df.write.nebula(config, nebulaWriteEdgeConfig).writeEdges()

The default write mode is insert, which can be changed to update or delete via withWriteMode configuration:

val config = NebulaConnectionConfig
  .builder()
  .withMetaAddress("127.0.0.1:9559")
  .withGraphAddress("127.0.0.1:9669")
  .build()
val nebulaWriteVertexConfig = WriteNebulaVertexConfig
  .builder()
  .withSpace("test")
  .withTag("person")
  .withVidField("id")
  .withVidAsProp(true)
  .withBatch(1000)
  .withWriteMode(WriteMode.UPDATE)
  .build()
df.write.nebula(config, nebulaWriteVertexConfig).writeVertices()

NebulaConnectionConfig is the configuration for connecting to the nebula graph, as described below.

Parameter	Required	Description
`withMetaAddress`	Yes	Specifies the IP addresses and ports of all Meta Services. Separate multiple addresses with commas. The format is `ip1:port1,ip2:port2,...`.
`withGraphAddress`	Yes	Specifies the IP addresses and ports of Graph Services. Separate multiple addresses with commas. The format is `ip1:port1,ip2:port2,...`.
`withConnectionRetry`	No	Number of retries that the NebulaGraph Java Client connected to the NebulaGraph. The default value is `1`.

WriteNebulaVertexConfig is the configuration of the write vertex, as described below.

Parameter	Required	Description
`withSpace`	Yes	NebulaGraph space name.
`withTag`	Yes	The Tag name that needs to be associated when a vertex is written.
`withVidField`	Yes	The column in the DataFrame as the vertex ID.
`withVidPolicy`	No	When writing the vertex ID, NebulaGraph use mapping function, supports HASH only. No mapping is performed by default.
`withVidAsProp`	No	Whether the column in the DataFrame that is the vertex ID is also written as an property. The default value is `false`. If set to `true`, make sure the Tag has the same property name as `VidField`.
`withUser`	No	NebulaGraph user name. If authentication is disabled, you do not need to configure the user name and password.
`withPasswd`	No	The password for the NebulaGraph user name.
`withBatch`	Yes	The number of rows of data written at a time. The default value is `1000`.
`withWriteMode`	No	Write mode. The optional values are `insert`, `update` and `delete`. The default value is `insert`.
`withDeleteEdge`	No	Whether to delete the related edges synchronously when deleting a vertex. The default value is `false`. It takes effect when `withWriteMode` is `delete`.

WriteNebulaEdgeConfig is the configuration of the write edge, as described below.

Parameter	Required	Description
`withSpace`	Yes	NebulaGraph space name.
`withEdge`	Yes	The Edge type name that needs to be associated when a edge is written.
`withSrcIdField`	Yes	The column in the DataFrame as the vertex ID.
`withSrcPolicy`	No	When writing the starting vertex ID, NebulaGraph use mapping function, supports HASH only. No mapping is performed by default.
`withDstIdField`	Yes	The column in the DataFrame that serves as the destination vertex.
`withDstPolicy`	No	When writing the destination vertex ID, NebulaGraph use mapping function, supports HASH only. No mapping is performed by default.
`withRankField`	No	The column in the DataFrame as the rank. Rank is not written by default.
`withSrcAsProperty`	No	Whether the column in the DataFrame that is the starting vertex is also written as an property. The default value is `false`. If set to `true`, make sure Edge type has the same property name as `SrcIdField`.
`withDstAsProperty`	No	Whether column that are destination vertex in the DataFrame are also written as property. The default value is `false`. If set to `true`, make sure Edge type has the same property name as `DstIdField`.
`withRankAsProperty`	No	Whether column in the DataFrame that is the rank is also written as property.The default value is `false`. If set to `true`, make sure Edge type has the same property name as `RankField`.
`withUser`	No	NebulaGraph user name. If authentication is disabled, you do not need to configure the user name and password.
`withPasswd`	No	The password for the NebulaGraph user name.
`withBatch`	Yes	The number of rows of data written at a time. The default value is `1000`.
`withWriteMode`	No	Write mode. The optional values are `insert`, `update` and `delete`. The default value is `insert`.

Last update: February 19, 2024