Versions Compared

Version Old Version 15 New Version 16
Changes made by

Yuri Au Yong

Yuri Au Yong

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

You open the GCP BigQuery agent configuration dialog from a workflow configuration: you can right-click the agent icon and select Configuration..., or double-click the agent icon

...

FieldDescription
GCP ProfileClick Browse to select a predefined GCP profile. The profile contains the credential information required to connect to the BigQuery in the GCP Project.
UDR Type

Type of UDR to populate the target BigQuery table.

DatasetSelect the target dataset.
Data Table

Select the target table name. Only the tables found within the selected dataset can be listed and chosen.

Info
titleInfo!

When creating the Data table, ensure that:

  • The table must have a Transaction ID column, dedicated to the agent's internal use for transaction safety purposes. The column could be named according to your preference  however it must be BigQuery INT64 data type (with alias INT, SMALLINT, INTEGER, BIGINT, TINYINT, BYTEINT). It must also not allow NULL. Take note of the name as you will have to create a similar column with the same name for the Batch Status table.


Info
titleInfo!

Due to the restriction on modifying rows in Data table, the Data table is not modified during Commit and Rollback workflow stages. This Restriction is explained further in 4.2 GCP BigQuery Agent Transaction Behavior.


Batch Status Table

A separate table to keep track of the transaction ID to allow the agent to better provide transaction safety in the event that the workflow aborts. A transaction ID is unique to each batch process, the agent will use the transaction ID to identify UDRs that belong to the batch when the data insertion into the table was interrupted.

Info
titleInfo!

When creating the Batch Status table, ensure that:

  • The table must have a Transaction ID column, dedicated to the agent's internal use for transaction safety purposes. The column must be named the same as the Transaction ID column created in the Data table and it must be in INT64. It must also not allow NULL.

  • The table must also have a Status column for the agent to update the status of the transaction. This column must also be BigQuery INT64 data type (with alias INT, SMALLINT, INTEGER, BIGINT, TINYINT, BYTEINT) and must be named status.


See the example below on creating a Batch Status table:

Info
titleExample - Batch Status Table

The following DDL query is used in the BigQuery Query Editor to create a table under the user_analytics Dataset with the table named batch_status.

Code Block
 CREATE TABLE user_analytics.batch_status
 (
   transactionID INT64 NOT NULL OPTIONS(description="The transaction ID field"),
   status INT64 OPTIONS(description="The status field"),
 )



Info
titleInfo!

The Batch Status table is updated with status code during the Commit and Rollback workflow stages. The Batch Status, Status column may contain the following codes:

CodeDescription
100The UDR's are currently being inserted into the Data table.
0The workflow successfully inserted all the UDRs into the Data table.
1The workflow has aborted or has been manually stopped and the UDRs insertion has been cancelled.
2The Workflow entered rollback stage, possibly due to errors in previous workflow run.
-1This status indicates an unknown issue and requires further investigation into the cause.



Transaction ID ColumnSelect the column that is designated for the transaction ID from the Batch Status Table and Data Table. The transaction ID has to be present in both Data and Batch Status table with the same name, for the transaction ID to be available for selection.
Batch Size

The amount of UDR to be inserted into the target data table for each batch process. This corresponds to the rows per insert request into the target Data table. The default value is 500 records per batch transaction.

Note
titleNote!

If you need to increase the Batch Size, you will need to refer to the GCP documentation on the maximum limit of Bigquery streaming inserts. 


Concurrent Requests

The amount of connections the agent will open to the target data table to insert the UDRs.

Note
titleNote!

The order of insertion of the UDR data into the BigQuery table is done out of sequence and will appear as such in the target table. Sorting will have to be done from BigQuery itself.


...

FieldDescription
Refresh

Updates the table with all the columns from the selected table.

Note
titleNote!
Potential changes in the database table will not be visible until the Dataset, Data Table, Batch Status Table and Transaction ID Column have been properly configured.

If rows already exist in the table, the refresh operation preserves the configuration for all rows with a corresponding column. Thus, if a table has been extended with a new column, the old column configurations are left untouched and the new column appears when  Refresh is selected.

Field Name

Displays a list of all columns for the selected Data table, except the Transaction ID column.

Field Type

Displays the BigQuery data type for each column as declared in the BigQuery table.

Info
titleInfo!

BigQuery data types that are currently supported are: BIGNUMERIC, BOOLEAN, DATE, DATETIME, TIME, TIMESTAMP, FLOAT64, NUMERIC, GEOGRAPHY, STRING and INT64.


UDR FieldAllows you to map the Field Name of the Data table with the UDR Field from the UDR Type configured from the Target tab. Only valid UDR Field types will be available for selection in the Column Assignment dialog based on the Field Type of the BigQuery Column. You can refer to the BigQuery to valid UDR data types mappings in the table below.

...