Synchronize data from an Oracle database to a DataHub instance

Alibaba Cloud DataHub is a streaming data processing platform for publishing, subscribing to, and distributing streaming data, enabling easy analysis and application based on streaming data.

Limits

• OceanBase Migration Service (OMS) does not support incremental synchronization of a table if all the columns in the table that are of the LOB type.

• The Shard Columns parameter must be set for tables without a primary key in the Oracle database for successful synchronization to the DataHub instance.

• OMS cannot parse the actual values of the generated columns used in the Oracle database. Therefore, when data is synchronized to the DataHub instance, the corresponding values are NULL.

• Data source identifiers, user accounts, and tags must be globally unique in OMS.

• OMS supports only the synchronization of objects whose database name or table name is an ASCII string without special characters. The special characters are . | \ " ' ` ( ) = ; / & \n

• During the incremental synchronization from an Oracle 12c database or later to a DataHub instance, if you want to add or change columns, the table names and column names cannot exceed 30 bytes in length.

  If you want the database to support table names and column names of more than 30 bytes in length, specify the ENABLE_GOLDENGATE_REPLICATION parameter of the Oracle database as the SYS user, and set deliver2store.logminer.need_check_object_length of Oracle Store to false.

   * Execute the following statement to set ENABLE_GOLDENGATE_REPLICATION: For a Real Application Cluster (RAC) environment, set this parameter for each node. If the Oracle database is in Active Data Guard (ADG) mode, set this parameter in the ADG source database.
  
      ```sql
      ALTER SYSTEM SET ENABLE_GOLDENGATE_REPLICATION=true SCOPE=BOTH;
      ```
  
   * Execute the following statement to query the `ENABLE_GOLDENGATE_REPLICATION` parameter:
  
      ```sql
      SELECT K.KSPPINM,V.KSPPSTVL FROM SYS.X$KSPPI K,SYS.X$KSPPSV V WHERE K.INDX=V.INDX AND UPPER(K.KSPPINM) = 'ENABLE_GOLDENGATE_REPLICATION';
      ```
  

Usage notes

• When the Oracle database runs in single standby or primary/standby mode, if the number of running instances in the primary database is different from that in the standby database, the incremental logs of some instances may be missed during data pulling. In this case, you must manually set an Oracle store parameter to specify the instances whose incremental logs are to be pulled from the standby database during incremental synchronization. Perform the following steps:

  1. After an Oracle store starts, immediately stop it.

  2. On the Update Configuration page of the store, add the deliver2store.logminer.instance_threads parameter to specify the instances whose logs are to be pulled.

     Separate multiple threads with vertical bars (|), such as 1|2|3. For more information, see Update the configurations of a store.

  3. After you specify the parameter, restart the Oracle store.

  4. 5 minutes later, run the grep 'log entries from' connector/connector.log command to view the instances whose logs are pulled. In the command output, the thread field indicates an instance whose logs are pulled.

• To perform incremental migration for an Oracle database, make sure that the size of each single archived file in the Oracle database is smaller than 2 GB. If the size of a single archived file is too large, the following risks may occur:

  • The pulling time increases more than linearly with the increase in file size.

  • When the Oracle database runs in single standby or primary/standby mode, incremental data is pulled from the standby database and only archived files can be pulled. An archived file is pulled only after it is generated. A larger size of the archived file indicates a high latency before this archived file is processed and a longer time consumed in processing this large archived file.

  • When pulling data with the same degree of parallelism (DOP), Oracle Store needs more memory to read large archived files.

• The archived files of the Oracle database are retained for more than 2 days. This avoids recovery failures due to the absence of archived files when the number of archived files sharply increases during a certain period or a processing exception occurs in the Oracle store.

• If the clocks are not synchronized between the nodes or between the client and a server, the latency of incremental synchronization may be negative.

• When data transmission is resumed for a project, some data (transmitted within the last minute) may be duplicated in the DataHub instance. Therefore, data deduplication is required in downstream applications.

• We recommend that you select no more than 15,000 database objects for a project.

  If a table contains LOB fields or contains more than 500 columns, we recommend that you create a dedicated project for the table and set the JVM parameters of related components as needed. For example, set the limitator.select.batch.max parameter for the full verification component, the sourceBatchSize parameter for the full import component, and the sourceBatchSize parameter for the incremental synchronization component.

  Execute the following statement to query tables that contain LOB fields: SELECT DISTINCT(TABLE_NAME) FROM ALL_TAB_COLUMNS WHERE DATA_TYPE IN ('BLOB', 'CLOB', 'NCLOB') AND OWNER = XXX;.

• If LogMiner generates invalid time data, such as 13621-11-11 11:32:08, during data synchronization, the Oracle store generates an error.

  In this case, you can perform the following operations: Choose OPS & Monitoring > Components > Store. On the page that appears, click Update for the target store, add the deliver2store.logminer.replace_invalid_date parameter, and set it to true. Then skip the data in the data synchronization project.

  If users are not sensitive to DATE data, you can set the deliver2store.logminer.replace_invalid_date parameter to TRUE to enable the reader to continue running. When the Oracle store generates data, it converts abnormal DATE data into the date when the logs were written to the disk.

Data type mappings

Check and modify the configurations of the source Oracle database

• Check the character set configurations

  OMS allows you to synchronize data from the source Oracle database based on the AL32UTF8, AL16UTF16, ZHS16GBK, or GB18030 character set.

• Check and modify the system configurations of the Oracle instance

  1. Enable archivelog and supplemental_log for the source Oracle database.

  2. In the Oracle database, perform the following operations as the sys user.

     Execute the following statement to check whether log_mode is set to archivelog and supplemental_log parameters are set to yes or implicit:

     select log_mode, supplemental_log_data_pk, supplemental_log_data_ui, supplemental_log_data_min from v$database;
     

     If not, use the following syntax to modify the configuration of the Oracle database:

     ALTER DATABASE ADD SUPPLEMENTAL LOG DATA;
     ALTER DATABASE ADD SUPPLEMENTAL LOG DATA(PRIMARY KEY) COLUMNS;
     ALTER DATABASE ADD SUPPLEMENTAL LOG DATA(UNIQUE) COLUMNS;
     

  3. Restart the Oracle database.

Supplemental properties

If you manually create a topic, add the following properties to the DataHub schema before you start a data synchronization project. If OMS automatically creates a topic and synchronizes the schema, OMS automatically adds the following properties.

Procedure

1. Create a data synchronization project.

   1. Log on to the OMS console.

   2. In the left-side navigation pane, click Data Synchronization.

   3. On the Data Synchronization page, click Create Synchronization Project in the upper-right corner.

2. On the Select Source and Destination page, specify the following parameters.

   Parameter	Description
   Synchronization Project Name	We recommend that you set it to a combination of digits and letters. It must not contain any spaces and cannot exceed 64 characters in length.
   Labels	Click the field and select a target tag from the drop-down list. You can click Manage Tags to create, modify, and delete tags. For more information, see Manage data synchronization projects by using tags.
   Source	If you have created an Oracle data source, select it from the drop-down list. Otherwise, click Add Data Source in the drop-down list to create one in the dialog box on the right side. For more information about the parameters, see Create an Oracle data source.
   Destination	If you have created a DataHub data source, select it from the drop-down list. Otherwise, click Add Data Source in the drop-down list to create one in the dialog box on the right side. For more information about parameters, see Create a DataHub data source.

3. Click Next. On the Select Synchronization Type page, select the synchronization type for the current data synchronization project.

   The supported synchronization types include Schema Synchronization and Incremental Synchronization. Schema Synchronization creates a topic. Incremental Synchronization supports only the DML Synchronization option. The supported DML operations are Insert, Delete, and Update. Select the options based on your business needs. For more information, see DML filtering.

4. Click Next. On the Select Synchronization Objects page, select the type and range of topics to be synchronized.

   Available topic types are Tuple and Blob. Tuple topics contain records that are similar to data records in databases. Each record contains multiple columns. You can only write a block of binary data as a record to a Blob topic. The data are Base64-encoded for transmission. For more information, visit the documentation center of DataHub.

   Select the type of topics to be synchronized and perform the following steps:

   1. In the left-side pane, select the objects to be synchronized.

      Notice

      The name of a table to be synchronized, as well as the names of columns in the table, must not contain Chinese characters.

   2. Click >.

   3. Select a mapping method.

      Notice

      When you set the topic type to tuple without selecting Schema Synchronization, you can only synchronize a single table to a single topic.

      • To synchronize a single table, select the mapping method as needed in the Map Object to Topic dialog box and click OK.

        If you do not select Schema Synchronization when you set the synchronization type and configuration, you can select only Existing Topics here. If you have selected Schema Synchronization when you set the synchronization type and configuration, you can select only one mapping method to create or select a topic.

        For example, if you selected Schema Synchronization, when you use both the Create Topic and Select Topic mapping methods or rename the topic, a precheck error will be returned due to option conflicts.

        Parameter	Description
        Create Topic	Enter the name of the new topic in the text box. The topic name can contain letters, digits, and underscores (_) and must start with a letter. It must not exceed 128 characters in length.
        Select Topic	OMS allows you to query DataHub topics. You can click Select Topic, and then find and select the topics to be synchronized from the Existing Topics drop-down list. You can also enter the name of an existing topic and select it after it appears.
        Batch Generate Topics	The format for generating topics in batches is: Topic_${Database Name}_${Table Name}.

        If you select Create Topic or Batch Generate Topics, after the schema migration succeeds, you can query the created topics on the DataHub side. By default, the number of data shards is 2 and the data expiration time is 7 days. These parameters cannot be modified. If the topics do not meet your business needs, you can create topics in the destination database as needed.

      • To synchronize multiple tables, click OK in the dialog box that appears.

        If you have selected a tuple topic and multiple tables without selecting Schema Synchronization, after you select a topic and click OK in the Map Object to Topic dialog box, multiple tables are displayed under the topic in the right pane, but only one table can be synchronized. Click Next. A prompt appears, indicating that only one-to-one mapping is supported between tuple topics and tables.

   4. Click OK.

   When you synchronize data from an Oracle database to a DataHub instance, OMS allows you to import objects from text data, set sharding columns for tables in the destination database, and remove a single object or all objects. Objects in the destination database are listed in the structure of Topic > Database > Table.

   Actions	Steps
   Import Objects	In the list on the right, click Import Objects in the upper-right corner. In the dialog box that appears, click OK. Notice This operation will overwrite previous selections. Proceed with caution. In the Import Synchronization Objects dialog box, import the objects to be synchronized. You can configure synchronization objects by importing a CSV file. For more information, see Download and import the settings of synchronization objects. Click Validate. After the validation succeeds, click OK.
   Change Topic	When the topic type is set to Blob, you can change topics for objects in the destination database. For more information, see Change a topic.
   Parameter	In the list on the right, move the pointer over the target object. Click Settings. Click the Shard Columns drop-down list and select the target sharding columns. You can select multiple fields as sharding columns. This parameter is optional. Unless otherwise specified, select the primary keys as sharding columns. If the primary keys are not load-balanced, select fields with unique identifiers and whose load is balanced as sharding columns to avoid potential performance issues. Sharding columns can be used for the following purposes: Load balancing: Threads used for sending messages can be recognized based on the sharding columns if the destination table supports concurrent writes. Orderliness: OMS ensures that messages are received in order if the values of the sharding columns are the same. The orderliness specifies the sequence of executing DML statements for a column. In the Select Columns section, select the columns to be synchronized. If you select all or no columns, OMS will synchronize all columns. Click OK.
   Remove/Remove All	During data mapping, OMS allows you to remove one or more selected objects to be migrated to the destination. Remove a single synchronization object In the list on the right of the selection section, move the pointer over the target object, and click Remove. The synchronization object is removed. Remove all synchronization objects In the list on the right of the selection section, click Remove All in the upper-right corner. In the dialog box that appears, click OK to remove all synchronization objects.

5. Click Next. On the Synchronization Options page, specify the following parameters.

   Parameter	Description
   Incremental Synchronization Start Timestamp	This parameter specifies to synchronize data after the synchronization start timestamp. The default value is the current system time. You can select a point in time or enter a timestamp. Notice You can select the current time or a point in time earlier than the current time. This parameter is closely related to the retention period of archived logs. Generally, you can start data synchronization from the current timestamp.
   Serialization Method	The message format for synchronizing data to a DataHub instance. Valid values: Default, Canal, Dataworks (version 2.0 supported), SharePlex, and DefaultExtendColumnType. For more information, see Data formats. Notice This parameter is available only when the topic type is set to Blob on the Select Synchronization Type page.
   Enable Intra-Transaction Sequence	Specifies whether to maintain order within a transaction. If you enable this feature, OMS marks each transaction sent to downstream devices. Notice This parameter is valid only for the SharePlex format and is intended for you to obtain the sequence numbers of the DML statements that form a transaction. For example, if a transaction contains 10 DML statements numbered from 1 to 10, OMS will deliver these statements to the destination database in the same order.
   Partitioning Rule	The rule for synchronizing data from the source database to a DataHub topic. Valid values: Hash and Table. Hash indicates that OMS uses a hash algorithm to select the shard of a DataHub topic based on the value of the primary key or sharding column. Table indicates that OMS delivers all data in a table to the same partition and uses the table name as the hash key.
   Business System Identification (Optional)	Identifies the source business system of data. The business system identifier consists of 1 to 20 characters.

6. Click Precheck.

   During the precheck, OMS detects the connection with the destination data source. If an error is returned during the precheck:

   • You can identify and troubleshoot the issue and then perform the precheck again.

   • You can click Skip in the Actions column of the precheck item with the error. A dialog box will be displayed, prompting the impact caused if you skip this error. If you want to continue, click OK in the dialog box.

7. Click Start Project. If you do not need to start the project now, click Save to go to the details page of the data synchronization project. You can start the project later as needed.

   OMS allows you to modify the synchronization objects when the data synchronization project is running. For more information, see View and modify synchronization objects. After a data synchronization project is started, the synchronization objects will be executed based on the selected synchronization type. For more information, see the "View synchronization details" section in the View details of a data synchronization project topic.

   If data access fails due to a network failure or the slow startup of processes, go to the project list or the project details page and click Restore.