Typical Scenario: Importing Data from a Relational Database to HBase

Scenario

Use Loader to import data from a relational database to HBase.

Prerequisites

• You have obtained the service username and password for creating a Loader job.
• You have had the permission to access the HBase tables or phoenix tables that are used during job execution.
• You have obtained the username and password of the relational database.
• No disk space alarm is reported, and the available disk space is sufficient for importing and exporting data.
• If a configured task requires the Yarn queue function, the user must be authorized with related Yarn queue permission.
• The user who configures a task must obtain execution permission on the task and obtain usage permission on the related connection of the task.
• Before the operation, perform the following steps:
  1. Obtain the JAR file of the relational database driver and save it to the following directory on the active and standby Loader nodes: ${BIGDATA_HOME}/FusionInsight_Porter_*/install/FusionInsight-Sqoop-1.99.3/FusionInsight-Sqoop-1.99.3/server/webapps/loader/WEB-INF/ext-lib.
  2. Run the following command on the active and standby nodes as user root to modify the permission:

     cd ${BIGDATA_HOME}/FusionInsight_Porter_*/install/FusionInsight-Sqoop-1.99.3/FusionInsight-Sqoop-1.99.3/server/webapps/loader/WEB-INF/ext-lib

     chown omm:wheel JAR file name

     chmod 600 JAR file name

  3. Log in to FusionInsight Manager and choose Cluster > Services > Loader. On the Dashboard tab page that is displayed, click More and select Restart Service. In the displayed dialog box, enter the administrator password to restart the Loader service.

Procedure

Configure basic job information.

1. Access the Loader web UI.
   1. Log in to FusionInsight Manager.
   2. Choose Cluster > Services > Loader.
   3. Click LoaderServer(Node name, Active). The Loader web UI is displayed.
      Figure 1 Loader web UI
      

2. Click New Job to go to the Basic Information page and set basic job information.
   Figure 2 Basic Information page
   
   1. Set Name to the name of the job.
   2. Set Type to Import.
   3. Set Group to the group to which the job belongs. No group is created by default. You need to click Add to create a group and click OK to save the created group.
   4. Set Queue to the Yarn queue that executes the job. The default value is root.default.
   5. Set Priority to the priority of the Yarn queue that executes the job. The default value is NORMAL. The options are VERY_LOW, LOW, NORMAL, HIGH, and VERY_HIGH.

3. In the Connection area, click Add to create a connection, set Connector to generic-jdbc-connector or a dedicated database connector (oracle-connector, oracle-partition-connector, or mysql-fastpath-connector), set connection parameters, and click Test to verify whether the connection is available. When "Test Success" is displayed, click OK.
   

   • For connection to relational databases, general database connectors (generic-jdbc-connector) or dedicated database connectors (oracle-connector, oracle-partition-connector, and mysql-fastpath-connector) are available. However, compared with general database connectors, dedicated database connectors perform better in data import and export because they are optimized for specific database types.
   • When mysql-fastpath-connector is used, the mysqldump and mysqlimport commands of MySQL must be available on NodeManager nodes, and the MySQL client version to which the two commands belong must be compatible with the MySQL server version. If the two commands are unavailable or the versions are incompatible, install the MySQL client applications and tools following the instructions at http://dev.mysql.com/doc/refman/5.7/en/linux-installation-rpm.html.

   Table 1 generic-jdbc-connector connection parameters

   Parameter	Description	Example Value
   Name	Name of a relational database connection	dbName
   JDBC Driver Class	Name of a Java database connectivity (JDBC) driver class	oracle.jdbc.driver.OracleDriver
   JDBC Connection String	JDBC connection string, in the format of jdbc:oracle:thin:@Database IP address:Database port number:Database name	jdbc:oracle:thin:@10.16.0.1:1521:oradb
   Username	Username for connecting to the database	omm
   Password	Password for connecting to the database	xxxx
   JDBC Connection Properties	JDBC connection attribute. Click Add to manually add the attribute. Name: connection attribute name Value: connection attribute value	Name: socketTimeout Value: 20

   Configure data source information.

4. Click Next. On the displayed From page, set the data source information.

   Table 2 Parameter description

   Parameter	Description	Example Value
   Schema name	Database schema name. This parameter exists in the Table name schema.	dbo
   Table name	Database table name. This parameter exists in the Table name schema.	test
   Table SQL statement	SQL statement for Loader to query data to be imported in Table SQL statement mode. The SQL statement requires the query condition WHERE ${CONDITIONS}. Without this condition, the SQL statement cannot be run properly. An example SQL statement is as follows: select * from TABLE WHERE A>B and ${CONDITIONS}. If Table column names is set, the column specified by Table column names will replace the column queried in the SQL statement. This parameter cannot be set when Schema name or Table name is set. NOTE: You can use macros to define SQL Where statements. For details, see Using Macro Definitions in Configuration Items.	select * from test where ${CONDITIONS}
   Table column names	Table columns whose content is to be imported by Loader. Use commas (,) to separate multiple fields. If the parameter is not set, all the columns are imported and the Select * order is used as the column location.	-
   Partition column name	Database table column based on which to-be-imported data is determined. This parameter is used for partitioning in a Map job. You are advised to configure the primary key field. NOTE: A partition column must have an index. If no index exists, do not specify a partition column. If a partition column without an index is specified, the database server disk I/O will be busy, the access of other services to the database will be affected, and the import will take a long period. In multiple fields with indexes, select the field that has the most discrete value as the partition column. A partition column that is not discrete may result in load imbalance when multiple MapReduce jobs are imported. The sorting rules of partition columns must be case-sensitive. Otherwise, data may be lost during data import. You are not advised to select fields of the float or double type for the partition column. Otherwise, the records containing the minimum and maximum values of the partition column may fail to be imported due to precision issues.	id
   Nulls in partition column	Indicates whether to process records whose values are null in database table columns. true: Records whose values are null are processed. false: Records whose values are not null are processed.	true
   Need partition column	Indicates whether to specify a partition column.	true

   Configure data transformation.

5. Click Next. On the displayed Transform page, set the transformation operations in the data transformation process. For details about how to select operators and set parameters, see Operator Help and Table 3.

   Table 3 Input and output parameters of the operator

   Input Type	Output Type
   Table Input	HBase Output

   In input, drag Table Input to the grid. In output, drag HBase Output to the grid. Use an arrow to connect Table Input to HBase Output.

   Set data storage information and execute the job.

6. Click Next. On the displayed To page, set Storage type to HBASE_BULKLOAD or HBASE_PUTLIST based on the actual situation.

   Table 4 Parameter description

   Storage Type	Applicable Scenario	Parameter	Description	Example Value
   HBASE_BULKLOAD	Large data volume	HBase Instance	HBase service instance that Loader selects from all available HBase service instances in the cluster. If the selected HBase service instance is not added to the cluster, the HBase job cannot be run properly.	HBase
   		Clear data before import	Indicates whether to clear data in the original table before importing data. True indicates clearing data and False indicates not to clear data. If you do not set this parameter, the original table is not cleared by default.	true
   		Extractors	Number of Maps that are started at the same time in a MapReduce task of a data configuration operation. The value must be less than or equal to 3000.	20
   		Extractor Size	HBase does not support this parameter. Please set Extractors.	-
   HBASE_PUTLIST	Small data volume	HBase Instance	HBase service instance that Loader selects from all available HBase service instances in the cluster. If the selected HBase service instance is not added to the cluster, the HBase job cannot be run properly.	HBase
   		Extractors	Number of Maps that are started at the same time in a MapReduce task of a data configuration operation. The value must be less than or equal to 3000.	true
   		Extractor Size	HBase does not support this parameter. Please set Extractors.	-

7. Click Save and Run to save and run the job.

   View the job execution result.

8. Go to the Loader web UI. When Status is Succeeded, the job is complete.
   Figure 3 Viewing job details