Teradata Consumer

Supported pipeline types:
  • Data Collector

The Teradata Consumer origin reads data from multiple Teradata Database tables in multiple schemas through a JDBC connection. The origin returns data as a map with column names and field values. With the Teradata Consumer origin, you can enable Teradata Fast Export to quickly retrieve large amounts of data.
Tip: Though you can also use the JDBC Multitable Consumer origin or the JDBC Query Consumer origin to read from Teradata Database tables, use the Teradata Consumer origin to enable Teradata Fast Export and retrieve large amounts of data quickly.

By default, the origin processes tables incrementally, using primary key columns or user-defined offset columns to track its progress. You can configure the origin to perform non-incremental processing to enable it to also process tables that do not have a key or offset column.

The origin can use multiple threads to enable parallel processing of tables and partitions. Use the Teradata Consumer origin to read multiple tables from one or more schemas in the same database. For example, you might use the origin to perform database replication.

When you configure the Teradata Consumer origin, you define groups of database tables to read. The origin generates SQL queries based on the table configurations that you define.

When you configure the origin, you specify connection information and custom JDBC configuration properties to determine how the origin connects to the database. You also configure the number of threads to use and the maximum number of queries to run per second.

When you define the table configuration for the groups of tables that you want to process, you can optionally override the default key column and specify the initial offset to use. You can enable non-incremental processing for tables without key or offset columns. You can configure the origin to perform multithreaded partition processing, multithreaded table processing, or use the default - a mix of both. When configuring partitions, you can configure the offset size, number of active partitions, and offset conditions.

You define the strategy that the origin uses to create each batch of data and the number of batches to create from each result set. You can configure advanced properties, such as the initial order to read from tables, connection related properties, and transaction isolation. And you can specify what the origin does when encountering an unsupported data type: convert the data to string or stop the pipeline.

When the pipeline stops, Teradata Consumer notes where it stops reading. When the pipeline starts again, Teradata Consumer continues processing from where it stopped by default. You can reset the origin to process all available data, using any initial offsets that you defined.

You can configure advanced connection properties. To use a JDBC version older than 4.0, you specify the driver class name and define a health check query.

The origin can generate events for an event stream. For more information about dataflow triggers and the event framework, see Dataflow Triggers Overview.

Before you use the Teradata Consumer origin, you must complete the prerequisite tasks, including installing the Teradata stage library. The Teradata stage library is an Enterprise stage library that is free for development purposes only. For information about purchasing the stage library for use in production, contact StreamSets.

Prerequisites

Before using the Teradata Consumer origin, complete the following prerequisites:

Install the Teradata Stage Library

You must install the Teradata stage library before using the Teradata Consumer origin.

The Teradata stage library is an Enterprise stage library that is free for development purposes only. For information about purchasing the stage library for use in production, contact StreamSets.

You can install Enterprise stage libraries using Package Manager for a tarball Data Collector installation or as custom stage libraries for a tarball, RPM, or Cloudera Manager Data Collector installation.

Supported Versions

The following table lists the versions of the Teradata Enterprise stage library to use with specific Data Collector versions:
Data Collector Version Supported Stage Library Version
Data Collector 3.8.x, 3.9.x, and 3.10.x Teradata Enterprise Library 1.0.1
Data Collector 3.7.x Teradata Enterprise Library 1.0.0

Installing with Package Manager

You can use Package Manager to install the Teradata Enterprise stage library on a tarball Data Collector installation.

  1. Click the Package Manager icon: .
  2. In the Navigation panel, click Enterprise Stage Libraries.
  3. Select Teradata Enterprise Library, then click the Install icon: .
  4. Read the StreamSets subscription terms of service. If you agree, select the checkbox and click Install.
    Data Collector installs the selected stage library.
  5. Restart Data Collector.

Installing as a Custom Stage Library

You can install the Teradata Enterprise stage library as a custom stage library on a tarball, RPM, or Cloudera Manager Data Collector installation.

  1. To download the stage library, go to the StreamSets Download Enterprise Connectors page.
    The web page displays the Enterprise stage libraries organized by release date, with the latest versions at the top of the page.
  2. Click the Enterprise stage library name and version that you want to download.
  3. In the Download Enterprise Connectors form, enter your name and contact information.
  4. Read the StreamSets subscription terms of service. If you agree, accept the terms of service and click Submit.
    The stage library downloads.
  5. Install and manage the Enterprise stage library as a custom stage library.
    For more information, see Custom Stage Libraries.

Installing the JDBC Driver

Before you use the Teradata Consumer origin, install the JDBC driver for Teradata. The origin cannot access the database until you install this driver.
  1. Download the Teradata JDBC driver from the Teradata website.

    The Teradata JDBC Driver consists of two .jar files.

  2. Install each .jar file as an external library for the Teradata Enterprise stage library.

For information about installing additional drivers, see Install External Libraries.

Note: StreamSets has tested the Teradata Consumer origin using Teradata Database release 16.20 with the Teradata JDBC Driver version 16.20.00.08.

Table Configuration

When you configure a Teradata Consumer origin, you define a table configuration for each group of tables that you want to read. A table configuration defines a group of tables with the same table name pattern, that are from one or more schemas with the same name pattern, and that have proper primary keys or the same user-defined offset columns.

You can define one or more table configurations.

For example, you can define one table configuration to replicate a database that has a proper primary key for each table. You simply enter the schema name and use the default table name pattern % which matches all tables in the schema.

Let's look at an example where you need to define more than one table configuration. Let's say that you want to copy tables from Teradata Database to an HBase cluster. The SALES schema contains ten tables, but you want to copy only the following four tables:
  • store_a
  • store_b
  • store_c
  • customers

The three store tables use orderID as the primary key. You want to override the primary key for the customers table, and so need to define customerID as the offset column for that table. You want to read all available data in the tables, so do not need to define an initial offset value.

You define one table configuration as follows so that the origin can read the three store tables:

  • Schema - SALES
  • Table Name Pattern - store%

Then you define the second table configuration as follows so that the origin can read the customers table:

  • Schema - SALES
  • Table Name Pattern - customers
  • Override Offset Columns - enabled
  • Offset Columns - customerID

Let's take a closer look at the schema and table name patterns and offset properties that you define in a table configuration.

Schema and Table Name Patterns

You define the group of tables that the Teradata Consumer origin reads by defining schema and table name patterns for the table configuration. The origin reads all tables whose names match the table pattern in the schemas whose names match the schema pattern.

The schema and table name patterns use the SQL LIKE syntax. For example, the LIKE syntax uses the percentage wildcard (%) to represent any string of zero or more characters. The schema name pattern st% matches schemas whose names start with st. The default table name pattern % matches all tables in the specified schemas.

For more information about valid patterns for the SQL LIKE syntax, see https://msdn.microsoft.com/en-us/library/ms179859.aspx.

You can optionally define a schema or table exclusion pattern to exclude some schemas or tables from being read. The schema and table exclusion patterns use a Java-based regular expression, or regex. For more information about using regular expressions with Data Collector, see Regular Expressions Overview.

For example, let's say that you want to read all tables in the US_WEST and US_EAST schemas except for tables that start with dept. You enter the following schema name, table name, and table exclusion pattern:
  • Schema - US%
  • Table Name Pattern - %
  • Table Exclusion Pattern - dept.*

Since you do not need to exclude any schemas, you simply leave the schema exclusion pattern empty.

Or, let's say that you want to read all tables from all schemas, except for the sys and system schemas. You enter the following schema name, table name, and schema exclusion pattern, and leave the table exclusion pattern blank:
  • Schema - %
  • Table Name Pattern - %
  • Schema Exclusion Pattern - sys|system

Offset Column and Value

The Teradata Consumer origin uses an offset column and initial offset value to determine where to start reading data within tables and partitions.

By default, the origin uses the primary key of the tables as the offset column and uses no initial offset value. When you use multithreaded table processing and the table has a composite primary key, the origin uses each primary key as an offset column. You cannot use composite keys with multithreaded partition processing.

By default, the origin reads all available data from each table when you start the pipeline. The origin generates SQL queries using the following syntax when you start the pipeline:
SELECT * FROM <table> ORDER BY <offset column_1>, <offset column_2>, ...

Where <offset column_n> represents each primary key of the table, such as when the table has a composite primary key. When you restart the pipeline or when the origin switches back to a previously read table, the origin adds a WHERE clause to the SQL query to continue reading from the last saved offset.

To use this default behavior, you do not need to configure any of the offset properties.

You can make the following changes to how the origin handles offset columns and initial offset values:
Override the primary key as the offset column
You can override the primary key and define another offset column or columns. Or if the table doesn’t have a primary key, you can define the offset column or columns to use.
Important: As a best practice, a user-defined offset column should be an incremental and unique column. If the column is not unique - that is, multiple rows can have the same value for this column - there is a potential for data loss upon pipeline restart. For details, see Multiple Offset Value Handling.
Having an index on this column is strongly encouraged since the underlying query uses an ORDER BY and inequality operators on this column.
Define an initial offset value
The initial offset value is a value within the offset column where you want the origin to start reading. When you define an initial offset value, you must first enter the offset column name and then the value. If you are using the default primary key as the offset column, enter the name of the primary key.
If you define an initial offset value for a single offset column, the origin generates SQL queries using the following syntax:
SELECT * FROM <table> ORDER BY <offset column> WHERE <offset column> > ${offset}
If you defined multiple offset columns, you must define an initial offset value for each column, in the same order that the columns are defined. The origin uses the initial offset values of all columns to determine where to start reading data. For example, you override the primary key with the following offset columns: p1, p2, p3 and define an initial offset value for each column. The origin generates SQL queries using the following syntax:
SELECT * FROM <table> ORDER BY p1, p2, p3 WHERE (p1 > ${offset1}) OR (p1 = ${offset1} AND p2 > ${offset2}) OR (p1 = ${offset1} AND p2 = ${offset2} AND p3 > ${offset3})
Note: Data Collector stores offsets for Datetime columns as Long values. For offset columns with a Datetime data type, enter the initial value as a Long value. You can use the time functions to transform a Datetime value to a Long value. For example, the following expression converts a date entered as a String to a Date object, and then to a Long value:
${time:dateTimeToMilliseconds(time:extractDateFromString('2017-05-01 20:15:30.915','yyyy-MM-dd HH:mm:ss.SSS'))} 
Define additional offset column conditions
You can use the expression language to define additional conditions that the origin uses to determine where to start reading data. The origin adds the defined condition to the WHERE clause of the SQL query.
You can use the offset:column function in the condition to access an offset column by position. For example, if you have a table with offset columns p1 and p2, then offset:column(0) returns the value of p1 while offset:column(1) returns the value of p2.
Let's say that you defined a transaction_time column as the offset column. While the origin reads the table, multiple active transactions are being written to the table with the current timestamp for the transaction_time column. When the origin finishes reading the first record with the current timestamp, the origin continues reading with the next offset and skips some rows with the current timestamp. You can enter the following offset column condition to ensure that the origin reads from all offset columns with a timestamp less than the current time:
${offset:column(0)} < ${time:now()}
If your database requires the datetime in a specific format, you can use the time:extractStringFromDate function to specify the format. For example:
${offset:column(0)} < '${time:extractStringFromDate(time:now(), "yyyy-MM-dd HH:mm:ss")}'

Reading from Views

The Teradata Consumer origin can read from views in addition to tables.

The origin reads from all tables and views that are included in the defined table configurations. If a table configuration includes views that you do not want to read, simply exclude them from the configuration.

Use the origin to read from simple views that select data from a single table.

We do not recommend using the origin to read from complex views that combine data from two or more tables using joins. If the origin reads from complex views, it runs multiple queries in parallel which can cause a heavy workload on the database.

Multithreaded Processing Modes

The Teradata Consumer origin performs parallel processing and enables the creation of a multithreaded pipeline. The origin can use multiple threads to process entire tables or partitions within tables.

By default, the origin performs multithreaded partition processing for the tables that fulfill the partition processing requirements, and performs multithreaded table processing for all other tables. When using the default behavior, the origin notes the tables that allow partition processing in the Data Collector log. When needed, you can configure the origin to require partition processing for all tables or to perform only table processing. You can also allow the single-threaded non-incremental processing of tables when needed.

The origin provides the following multithreaded processing modes:
  • Multithreaded table processing - The origin can use up to one thread per table. Can process tables with multiple offset columns.
  • Multithreaded partition processing - The origin can use up to one thread per table partition. Use to process larger volumes of data than multithreaded table processing.

    Multithreaded partition processing requires a single primary key or user-defined offset column of a supported data type, and additional details for partition creation. Tables with composite keys or a key or user-defined offset column of an unsupported data type cannot be partitioned.

When you configure the origin, you specify the tables to process and the multithreaded partition processing mode to use for each set of tables:
  • Off - Use to perform multithreaded table processing.

    Can be used to perform non-incremental loads of tables without key or offset columns, when enabled.

  • On (Best Effort) - Use to perform partition processing where possible and allow multithreaded table processing for tables with multiple key or offset columns.

    Can be used to perform non-incremental loads of tables without key or offset columns, when enabled.

  • On (Required) - Use to perform partition processing for all specified tables.

    Does not allow performing other types of processing for tables that do not meet the partition processing requirements.

Multithreaded Table Processing

When performing multithreaded table processing, the Teradata Consumer origin retrieves the list of tables defined in the table configuration when you start the pipeline. The origin then uses multiple concurrent threads based on the Number of Threads property. Each thread reads data from a single table, and each table can have a maximum of one thread read from it at a time.
Note: The Maximum Pool Size property on the Advanced tab defines the maximum number of connections the origin can make to the database. It must be equal to or greater than the value defined for the Number of Threads property.

As the pipeline runs, each thread connects to the origin system, creates a batch of data, and passes the batch to an available pipeline runner. A pipeline runner is a sourceless pipeline instance - an instance of the pipeline that includes all of the processors, executors, and destinations in the pipeline and handles all pipeline processing after the origin.

Each pipeline runner processes one batch at a time, just like a pipeline that runs on a single thread. When the flow of data slows, the pipeline runners wait idly until they are needed, generating an empty batch at regular intervals. You can configure the Runner Idle Time pipeline property to specify the interval or to opt out of empty batch generation.

Multithreaded pipelines preserve the order of records within each batch, just like a single-threaded pipeline. But since batches are processed by different pipeline runners, the order that batches are written to destinations is not ensured.

The order of batch processing depends on many factors. For more information, see Processing Queue.

For more information about multithreaded pipelines, see Multithreaded Pipeline Overview.

Example

Say you are reading from ten tables. You set the Number of Threads property to 5 and the Maximum Pool Size property to 6. When you start the pipeline, the origin retrieves the list of tables. The origin then creates five threads to read from the first five tables, and by default Data Collector creates a matching number of pipeline runners. Upon receiving data, a thread passes a batch to each of the pipeline runners for processing.

At any given moment, the five pipeline runners can each process a batch, so this multithreaded pipeline processes up to five batches at a time. When incoming data slows, the pipeline runners sit idle, available for use as soon as the data flow increases.

Multithreaded Partition Processing

By default, the Teradata Consumer origin performs multithreaded partition processing for all tables that meet the partition processing requirements, and performs table processing for all other tables.

To perform multithreaded processing of partitions within a table, you enable partition processing in the table configuration, then specify the partition size and the maximum number of partitions to use. Limiting the number of partitions also limits the number of threads that can be dedicated to processing data in the table.

When you configure a set of tables for unlimited partitions, the origin creates up to twice as many partitions as the pipeline thread count. For example, if you have 5 threads, the table can have up to 10 partitions.

Similar to multithreaded table processing, each thread reads data from a single partition, and each partition can have a maximum of one thread read from it at a time.

When processing partitions, the processing order depends on many factors. For a full description, see Processing Queue.

Partition Processing Requirements

To perform multithreaded partition processing for a table, the table must meet the following requirements:

Single key or offset column
The table must have a single primary key or user-defined offset column. Performing multithreaded partition processing on a table with composite keys generates an error and stops the pipeline.
If a table does not have a primary key column, you can use the Override Offset Columns property to specify a valid offset column to use. Having an ascending index on the offset column is strongly encouraged since the underlying query uses an ORDER BY and inequality operators on this column.
Numeric data type
To use partition processing, the primary key or user-defined offset column must have a numeric data type that allows arithmetic partitioning.
The key or offset column must be one of the following data types:
  • Integer-based: Integer, Smallint, Tinyint
  • Long-based: Bigint, Date, Time, Timestamp
  • Float-based: Float, Real
  • Double-based: Double
  • Precision-based: Decimal, Numeric

Multiple Offset Value Handling

When processing partitions, Teradata Consumer origin allows processing multiple records with the same offset value. For example, the origin can process multiple records with the same timestamp in a transaction_date offset column.

Warning: When processing multiple records with the same offset value, records can be dropped if you stop the pipeline when the origin is processing a series of records with the same offset value.

When you stop the pipeline as the origin is processing a series of records with the same offset value, the origin notes the offset. Then, when you restart the pipeline, it starts with a record with the next logical offset value, skipping any unprocessed records that use the same last-saved offset.

For example, say you specified a datetime column as a user-defined offset column, and five records in the table share the same datetime value. Now say you happen to stop the pipeline after it processes the second record. The pipeline stores the datetime value as the offset where it stopped. When you restart the pipeline, processing begins with the next datetime value, skipping the three unprocessed records with the last-saved offset value.

Best Effort: Processing Non-Compliant Tables

To process tables in a table configuration that might not meet the partition processing requirements, you can use the On (Best Effort) option when you configure the Multithreaded Partition Processing mode property.

When you select the best effort option, the origin performs multithreaded partition processing for all tables that meet the partition processing requirements. The origin performs multithreaded table processing for tables that include multiple key or offset columns. And if you enable non-incremental processing, the origin can also process all tables that do not include key or offset columns.

Non-Incremental Processing

You can configure the Teradata Consumer origin to perform non-incremental processing for tables with no primary keys or user-defined offset columns. By default, the origin performs incremental processing and does not process tables without a key or offset column.

You can enable non-incremental processing for the set of tables defined in a table configuration.

Note: When enabling non-incremental processing for a table without a key or offset column, you cannot require multithreaded partition processing for the table configuration. That is, you cannot run the pipeline with the Multithreaded Partition Processing Mode property set to On (Required).

Use On (Best Effort) or Off to perform non-incremental processing of the table. With either option selected, the table is processed using a single thread, like multithreaded table processing.

When you enable non-incremental processing, the origin processes any table without a key or offset column as follows:
  • The origin uses a single thread to process all available data in the table.
  • After processing all available data, the origin notes that the table has been processed as an offset. So, if you stop and restart the pipeline after the origin completes all processing, the origin does not reprocess the table.

    If you want to reprocess data in the table, you can reset the origin before restarting the pipeline. This resets the origin for all tables that the origin processes.

  • If the pipeline stops while the origin is still processing available data, when the pipeline restarts, the origin reprocesses the entire table. This occurs because the table has no key or offset column to allow for tracking progress.

For example, say you configure the origin to use five threads and process a set of tables that includes a table with no key or offset column. To process data in this table, you enable the Enable Non-Incremental Load table configuration property. You also set Multithreaded Partition Processing Mode to On (Best Effort) to allow the origin to use multithreaded partition processing when possible and allow both non-incremental processing and multithreaded table processing when needed.

When you start the pipeline, the origin allocates one thread to the table that requires non-incremental processing. It processes the table data using multithreaded table processing until all data is processed. When the thread completes processing all available data, the origin notes this as part of the offset, and the thread becomes available to process data from other tables. In the meantime, the four other threads process data from the rest of the tables using multithreaded partition processing when possible.

Batch Strategy

You can specify the batch strategy to use when processing data. The batch strategy behaves differently depending on whether you use multithreaded table processing or multithreaded partition processing. The behavior can also be affected by the Batches from Result Set property.

Process All Available Rows

The Process All Available Rows from the Table batch strategy differs slightly depending on whether the origin is processing full tables or partitions within a table.
Multithreaded table processing

When the origin performs multithreaded table processing for all tables, each thread creates multiple batches of data from one table, until all available rows are read from that table.

The thread runs one SQL query for all batches created from the table. Then, the thread switches to the next available table, running another SQL query to read all available rows from that table.

For example, let's say the origin has batch size of 100 and uses two concurrent threads to read from four tables, each of which contains 1,000 rows. The first thread runs a SQL query to create 10 batches of 100 rows each from table1, while the second thread uses the same strategy to read data from table2.

When table1 and table2 are fully read, the threads switch to table3 and table4 and complete the same process. When the first thread finishes reading from table3, the thread switches back to the next available table to read all available data from the last saved offset.

The number of threads that can process the tables is limited by the Number of Threads property for the origin.

When the tables being processed use both table and partition processing, the threads query the partitions as described below. For details on how the tables and partitions rotate through the processing queue, see Processing Queue.

Multithreaded partition processing

Multithreaded partition processing is similar to multithreaded table processing, except that it works at a partition level.

Each thread creates multiple batches of data from one partition. The number of batches created and processed at one time is based on the Batches from Result Set property.

Each thread runs one SQL query for the batches to be created from the partition. Then, the thread switches to the next available partition, running another SQL query.

For example, if you set the Batches from Result Set property to 3, a thread runs a query to create 3 batches of data from the partition being processed. When the thread completes processing the three batches, the thread becomes available to process the next partition or table in the processing queue.

The number of threads that can process partitions for each table is limited by the Number of Threads property for the origin and the Max Active Partitions table property.

For details on how the tables and partitions rotate through the processing queue, see Processing Queue.

Switch Tables

The Switch Tables batch strategy differs greatly depending on whether the origin performs full table or partition processing. The number of batches created and processed at one time is based on the Batches from Result Set property.
Multithreaded table processing

When the origin performs multithreaded table processing for all tables, each thread creates a set of batches from one table, and then switches to the next available table to create the next set of batches.

The thread runs an initial SQL query to create the first set of batches from the table. The database caches the remaining rows in a result set in the database for the same thread to access again, and then the thread switches to the next available table. A table is available in the following situations:
  • The table does not have an open result set cache. In this case, the thread runs an initial SQL query to create the first batch, caching the remaining rows in a result set in the database.
  • The table has an open result set cache created by that same thread. In this case, the thread creates the batch from the result set cache in the database rather than running another SQL query.

A table is not available when the table has an open result set cache created by another thread. No other thread can read from that table until the result set is closed.

When you configure a switch table strategy, define the result set cache size and the number of batches that a thread can create from the result set. After a thread creates the configured number of batches, a different thread can read from the table.
Note: By default, the origin instructs the database to cache an unlimited number of result sets. A thread can create an unlimited number of batches from that result set.

For example, let's say an origin has a batch size of 100 and uses two concurrent threads and to read from four tables, each of which contains 10,000 rows. You set the result set cache size to 500 and set the number of batches read from the result set to 5.

Thread1 runs an SQL query on table1, which returns all 10,000 rows. The thread creates a batch when it reads the first 100 rows. The next 400 rows are cached as a result set in the database. Since thread2 is similarly processing table2, thread1 switches to the next available table, table3, and repeats the same process. After creating a batch from table3, thread1 switches back to table1 and retrieves the next batch of rows from the result set that it previously cached in the database.

After thread1 creates five batches using the result set cache for table1. Thread1 then switches to the next available table. A different thread runs an SQL query to read additional rows from table1, beginning from the last saved offset.

When the tables being processed use both table and partition processing, the threads query the partitions as described below. For details on how the tables and partitions rotate through the processing queue, see Processing Queue.

Multithreaded partition processing

Multithreaded partition processing is similar to multithreaded table processing, with a twist - each thread creates a set of batches from one partition for a table, then all partitions from the same table are moved to the end of the processing queue. This allows the origin to switch to the next available table.

The behavior around caching the result set and the number of batches to process from the result set is the same, but at a partition level.

For examples of how tables and partitions rotate through the processing queue, see Processing Queue.

Initial Table Order Strategy

You can define the initial order that the origin uses to read the tables.

Define one of the following initial table order strategies:
None
Reads the tables in the order that they are listed in the database.
Alphabetical
Reads the tables in alphabetical order.
Referential Constraints
Reads the tables based on the dependencies between the tables. The origin reads the parent table first, and then reads the child tables that refer to the parent table with a foreign key.
You cannot use the referential constraints order when the tables to be read have a cyclic dependency. When the origin detects a cyclic dependency, the pipeline fails to validate with the following error:
JDBC_68 Tables referring to each other in a cyclic fashion.
Note that the referential constraints order can cause pipeline validation or initialization to slow down because the origin has to sort the tables before reading them.

The origin uses this table order only for the initial reading of the tables. When threads switch back to previously read tables, they read from the next available table, regardless of the defined order.

Processing Queue

The Teradata Consumer origin maintains a virtual queue to determine the data to process from different tables. The queue includes each table defined in the origin. When a table is to be processed by partition, multiple partitions for the table are added to the queue, limited by the Max Partitions property.

The origin rotates and reorganizes the queue based on the Per Batch Strategy property. And it processes data from the queue with the threads specified in the Number of Threads property and the Batches from Result Set property.

Below are some scenarios to help clarify how the queue works.

Multiple Tables, No Partition Processing

Say you have tables A, B, C and D that you configure for table processing. When you start the pipeline, the origin adds all tables to the queue. If configured, the Initial Table Order Strategy advanced property can affect the order. Let's assume we have no referential constraints and use alphabetical order:
A  B  C  D
When a thread becomes available, it processes data from the first table in the queue. The number of batches is based on the Batches from Result Set property. The processing of the tables depends on how you define the Per Batch Strategy property:
Process All Available Rows in the Table
With this batch strategy, threads do not start processing data in the next table until all available data is processed for the preceding table.

That is, table A remains at the front of the queue until all available data is processed. Then processing begins on table B. Table A moves to the back, remaining in the queue in case more data appears, as follows:

B  C  D  A  
Switch Tables
With this batch strategy, the order of the queue remains the same, but each thread performs a SQL query to create a set of batches based on the Batches from Result Set property. When it completes processing, it performs the same process with the next table in the queue.

After a thread takes a set of batches from table A, table A moves to the back of the queue:

B  C  D  A
The next thread takes a set of batches from table B. Then B moves to the back of the queue:
C  D  A  B  
So after processing 4 sets of batches, the queue looks like it did in the beginning:
A  B  C  D

Multiple Partitions, No Table Processing

Say you have table A, B, and C and all three tables are loaded up with lots of data to process. Tables A and B are configured with a maximum of 3 active partitions. And since table C has the largest volume of data, you allow an unlimited number of partitions. Again, let's use the alphabetical initial table ordering.

When you start the pipeline, each table is queued up with the maximum number of active partitions. And for table C, that means double the number of threads for the pipeline. So if we configure the pipeline for 4 threads, table C can have up to 8 partitions in the queue at any given time. So the initial queue looks like this:
A1  A2  A3  B1  B2  B3  C1  C2  C3  C4  C5  C6  C7  C8 
A partition remains in the queue until the origin confirms that there is no more data in the partition. When a thread becomes available, it creates a set of batches from the first partition of the first table in the queue. The number of batches is based on the Batches from Result Set property. The order of tables and partitions in the queue depends on how you define the Per Batch Strategy, as follows:
Process All Available Rows in the Table
When processing partitions, this batch strategy retains the original order of the queue, but rotates through the partitions as each thread processes a set of batches.
Note: In practice, this means that rows from subsequent tables can be processed before a previous table is completed, since available threads continue to pick up partitions from the queue.
For example, the four threads start processing on the first four partitions in the queue: A1, A2, A3, and B1. This puts B2 at the front of the queue, ready for the next available thread. And since the four partitions being processed have additional data to process, they go to the back of the queue. So processing of table B data begins before table A is fully processed.

The rest of the partitions remain in the original order as follows:

B2  B3  C1  C2  C3  C4  C5  C6  C7  C8  A1  A2  A3  B1
After the four threads process another four sets of batches, the queue looks like this:
C3  C4  C5  C6  C7  C8  A1  A2  A3  B1  B2  B3  C1  C2
Switch Tables
When processing partitions, this batch strategy forces all subsequent, consecutive partitions from the same table to the end of the queue each time a thread processes a set of batches from a partition.

Let's start again with the initial batch order:

A1  A2  A3  B1  B2  B3  C1  C2  C3  C4  C5  C6  C7  C8 
When a thread processes a set of batches from A1, it pushes the rest of the table A partitions to the end of the queue. This queues up the next table, table B, for processing. And since A1 still contains data, it takes the last spot, as follows:
B1  B2  B3  C1  C2  C3  C4  C5  C6  C7  C8  A2  A3  A1
As the second thread processes a set of batches from B1, the other B partitions are sent to the back, and since B1 still contains data, it takes the last spot as follows:
C1  C2  C3  C4  C5  C6  C7  C8  A2  A3  A1  B2  B3  B1
And as the third thread takes a set of batches from C1, the rest of the C partitions are pushed to the back, so the queue looks like this:
A2  A3  A1  B2  B3  B1  C2  C3  C4  C5  C6  C7  C8  C1

Both Partition and Table Processing

When processing a mix of full tables and partitioned tables, the queue basically behaves the same as when processing only partitions, with full tables being processed as partitioned tables with a single partition. Let's walk through it.

Say we have table A being processed without partitions, and table B configured with a maximum of 3 partitions, and table C with no limit. As in the example above, the pipeline has 4 threads to work with which allows 8 partitions to table C. Using the alphabetical initial table ordering, the initial queue looks like this:
A  B1  B2  B3  C1  C2  C3  C4  C5  C6  C7  C8 
When a thread becomes available, it processes a set of batches from the first table or partition in the queue. The number of batches is based on the Batches from Result Set property. The order of the queue depends on how you define the Per Batch Strategy, as follows:
Process All Available Rows in the Table
With this batch strategy, the queue remains in the basic initial order and rotates as each thread claims a set of batches from the next table or partition. The unpartitioned table A is processed like a table with a single partition.

Note that unpartitioned tables are not processed in full when they move to the front of the queue. For this behavior, configure all tables to be processed without partitions. Or, set the Batches from Result Set property to -1.

When the pipeline starts, the 4 threads process a set of batches from the A table and from partitions B1, B2, and B3. Since the table and partitions all still contain data, they then move to the end of the queue as follows:

C1  C2  C3  C4  C5  C6  C7  C8  A  B1  B2  B3 
As each thread completes processing, it processes a set of batches from the front of the queue. After each of the 4 threads takes another set of batches, the queue looks like this:
C5  C6  C7  C8  A  B1  B2  B3  C1  C2  C3  C4 
Switch Tables
When processing tables and partitions, this batch strategy forces all subsequent, consecutive partitions from the same table to the end of the queue. And it treats unpartitioned tables as a table with a single partition. As a result, the queue rotation is a simplified version of processing only partitioned tables.

So we have this initial order:

A  B1  B2  B3  C1  C2  C3  C4  C5  C6  C7  C8 
The first thread processes a set of batches from table A, and since there are no related partitions, it simply goes to the end of the queue:
B1  B2  B3  C1  C2  C3  C4  C5  C6  C7  C8  A
The second thread processes a set of batches from B1, pushes the rest of the table B partitions to the end of the queue, and B1 lands at the end because it contains more data to be processed:
C1  C2  C3  C4  C5  C6  C7  C8  A  B2  B3  B1
The third thread processes a set of batches from C1, pushes the rest of the table C partitions to the end, and C1 takes the last slot:
A  B2  B3  B1  C2  C3  C4  C5  C6  C7  C8  C1
And then the fourth thread processes another set of batches from table A, and moves A to the end of the queue:
B2  B3  B1  C2  C3  C4  C5  C6  C7  C8  C1  A

JDBC Header Attributes

The Teradata Consumer origin generates JDBC record header attributes that provide additional information about each record, such as the original data type of a field or the source tables for the record. The origin receives these details from the JDBC driver.

You can use the record:attribute or record:attributeOrDefault functions to access the information in the attributes. For more information about working with record header attributes, see Working with Header Attributes.

JDBC record header attributes include a "jdbc" prefix to differentiate the JDBC attributes from other record header attributes.

The origin can provide the following JDBC header attributes:
JDBC Header Attribute Description
jdbc.tables
Provides a comma-separated list of source tables for the fields in the record.
Note: Not all JDBC drivers provide this information.
jdbc.partition Provides the full offset key for the partition that produced the record
jdbc.threadNumber Provides the number of the thread that produced the record.
jdbc.<column name>.jdbcType Provides the numeric value of the original SQL data type for each field in the record. See the Java documentation for a list of the data types that correspond to numeric values.
jdbc.<column name>.precision Provides the original precision for all numeric and decimal fields.
jdbc.<column name>.scale Provides the original scale for all numeric and decimal fields.

Event Generation

The Teradata Consumer origin can generate events that you can use in an event stream. When you enable event generation, the origin generates an event when it completes processing the data returned by the specified queries for all tables. The origin also generates events when it completes processing the data returned from a table and the data returned from a schema.

Teradata Consumer events can be used in any logical way. For example:
  • With the Pipeline Finisher executor to stop the pipeline and transition the pipeline to a Finished state when the origin completes processing available data.

    When you restart a pipeline stopped by the Pipeline Finisher executor, the origin continues processing from the last-saved offset unless you reset the origin.

    For an example, see Case Study: Stop the Pipeline.

  • With the Email executor to send a custom email after receiving an event.

    For an example, see Case Study: Sending Email.

For more information about dataflow triggers and the event framework, see Dataflow Triggers Overview.

Event Record

Event records generated by Teradata Consumer origin have the following event-related record header attributes:
Record Header Attribute Description
sdc.event.type Event type. Uses the following type:
  • no-more-data - Generated when the origin completes processing all data returned by the queries for all tables.
  • schema-finished - Generated when the origin completes processing all rows within a schema.
  • table-finished - Generated when the origin completes processing all rows within a table.
sdc.event.version Integer that indicates the version of the event record type.
sdc.event.creation_timestamp Epoch timestamp when the stage created the event.

The Teradata Consumer origin can generate the following event record:

no-more-data
The Teradata Consumer origin generates a no-more-data event record when the origin completes processing all data returned by the queries for all tables.
You can configure the origin to delay the generation of the no-more-data event by a specified number of seconds. You might configure a delay to ensure that the schema-finished or table-finished events are generated and delivered to the pipeline before the no-more-data event record.

To use a delay, configure the No-more-data Event Generation Delay property.

The no-more-data event record generated by the origin has the sdc.event.type set to no-more-data and does not include any additional fields.

schema-finished
The Teradata Consumer origin generates a schema-finished event record when the origin completes processing all data within a schema.
The schema-finished event record has the following additional fields:
Event Record Field Description
schema The schema that has returned no remaining data to be processed.
tables A list of the tables within the schema that have no remaining data.
table-finished
The Teradata Consumer origin generates a table-finished event record when the origin completes processing all data within a table.
The table-finished event record has the following additional fields:
Event Record Field Description
schema The schema associated with the table that has no remaining data to be processed.
table The table that has no remaining data to be processed.

Configuring a Teradata Consumer Origin

Configure a Teradata Consumer origin to read data from multiple Teradata Database tables through a JDBC connection. Before you use the origin in a pipeline, complete the required prerequisites.

  1. In the Properties panel, on the General tab, configure the following properties:
    General Property Description
    Name Stage name.
    Description Optional description.
    Produce Events Generates event records when events occur. Use for event handling.
    On Record Error Error record handling for the stage:
    • Discard - Discards the record.
    • Send to Error - Sends the record to the pipeline for error handling.
    • Stop Pipeline - Stops the pipeline.
  2. On the JDBC tab, configure the following properties:
    JDBC Property Description
    JDBC Connection String Connection string used to connect to the database.

    The default value is jdbc:teradata://. Build your connection string from the default value. For example, you might enter:

    jdbc:teradata://<Host>/<ParameterName1=Value1,ParameterName2=Value2>

    Fast Export Enables Teradata Fast Export functionality. When enabled, the origin can quickly retrieve large amounts of data from a Teradata Database table or view, provided all Teradata data types in the source table or view support JDBC Fast Export. For more information, see the Teradata documentation.
    Use Credentials Enables entering credentials on the Credentials tab. Use when you do not include credentials in the JDBC connection string.
    Queries per Second Maximum number of queries to run in a second across all partitions and tables. Use 0 for no limit.

    Default is 10.

    Number of Threads Number of threads the origin generates and uses for multithreaded processing.

    Configure the Maximum Pool Size property on the Advanced tab to be equal to or greater than this value.

    Per Batch Strategy Strategy to create each batch of data:
    • Switch Tables - When performing only multithreaded table processing, each thread creates a batch of data from one table, and then switches to the next available table to create the next batch. Define the Result Set Cache Size and the Batches from Result Set properties when you configure a switch tables strategy.
    • Process All Available Rows From the Table - When performing only multithreaded table processing, each thread creates multiple batches of data from one table, until all available rows are read from that table.

    When performing multithreaded partition processing or a mix of table and partition processing, the behavior for each batch strategy is more complicated. For details, see Processing Queue.

    Max Batch Size (records) Maximum number of records to include in a batch.
    Batches from Result Set Number of batches to create from the result set. After a thread creates this number of batches, the database closes the result set and then another thread can read from the same table.

    Use a positive integer to set a limit on the number of batches created from the result set. Use -1 to opt out of this property.

    By default, the origin creates an unlimited number of batches from the result set, keeping the result set open as long as possible.

    Result Set Cache Size Number of result sets to cache in the database. Use a positive integer to set a limit on the number of cached result sets. Use -1 to opt out of this property.

    By default, the origin caches an unlimited number of result sets.

    Max Clob Size (characters) Maximum number of characters to be read in a Clob field. Larger data is truncated.
    Max Blob Size (bytes)

    Maximum number of bytes to be read in a Blob field.

    Number of Retries on SQL Error Number of times a thread tries to read a batch of data after receiving an SQL error. After a thread retries this number of times, the thread handles the error based on the error handling configured for the origin.

    Use to handle transient network or connection issues that prevent a thread from reading a batch of data.

    Default is 0.

    Data Time Zone Time zone to use to evaluate datetime-based offset column conditions.
    No-more-data Event Generation Delay (seconds) Number of seconds to delay generating the no-more-data event. Use to allow the specified number of seconds to pass to verify that no additional data arrives before generating the no-more-data event.
    Quote Character Quote character to use around schema, table, and column names in the query. Select the character used by the database to allow for lower case, mixed-case, or special characters in schema, table, or column names:
    • None - Uses no character around names in the query. For example: select * from mySchema.myTable order by myOffsetColumn.
    • Backtick - Uses a backtick around names in the query. For example: select * from `mySchema`.`myTable` order by `myOffsetColumn`.
    • Double Quotes - Uses double quotes around names in the query. For example: select * from "mySchema"."myTable" order by "myOffsetColumn".
    Convert Timestamp To String Enables the origin to write timestamps as string values rather than datetime values. Strings maintain the precision stored in the source database.

    When writing timestamps to Data Collector date or time data types that do not store nanoseconds, the origin stores any nanoseconds from the timestamp in a field attribute.

    Fetch Size Maximum number of rows to fetch and store in memory on the Data Collector machine. The size cannot be zero.

    Default is 1,000.

    For more information about configuring a fetch size, see your database documentation.

    Additional JDBC Configuration Properties Additional JDBC configuration properties to use. To add properties, click Add and define the JDBC property name and value.

    Use the property names and values as expected by JDBC.

  3. On the Tables tab, define one or more table configurations. Using simple or bulk edit mode, click the Add icon to define another table configuration.
    Configure the following properties for each table configuration:
    Tables Property Description
    Schema Pattern of the schema names to read for this table configuration. Use the SQL LIKE syntax to define the pattern.
    Table Name Pattern Pattern of the table names to read for this table configuration. Use the SQL LIKE syntax to define the pattern.

    Default is the percentage wildcard (%) which matches all tables in the schema.

    Table Exclusion Pattern Pattern of the table names to exclude from being read for this table configuration. Use a Java-based regular expression, or regex, to define the pattern.

    Leave empty if you do not need to exclude any tables.

    Schema Exclusion Pattern Pattern of the schema names to exclude from being read for this table configuration. Use a Java-based regular expression, or regex, to define the pattern.

    Leave empty if you do not need to exclude any schemas.

    Override Offset Columns Determines whether to use the primary keys or other columns as the offset columns for this table configuration.

    Select to override the primary keys and define other offset columns. Clear to use existing primary keys as the offset columns.

    To perform multithreaded partition processing on a table with multiple key columns or a key column with unsupported data types, select this option and specify a valid offset column. For more information about partition processing requirements, see Partition Processing Requirements.

    Offset Columns Offset columns to use.

    As a best practice, an offset column should be an incremental and unique column that does not contain null values. Having an index on this column is strongly encouraged since the underlying query uses an ORDER BY and inequality operators on this column.

    Initial Offset Offset value to use for this table configuration when the pipeline starts. Enter the primary key name or offset column name and value. For Datetime columns, enter a Long value.

    When you define multiple offset columns, you must define an initial offset value for each column, in the same order that the columns are defined.

    Enable Non-Incremental Load Enables non-incremental processing of tables that do not include a primary key or offset column. Do not use when requiring multithreaded partition processing.
    Multithreaded Partition Processing Mode Determines how the origin performs multithreaded processing. Select one of the following options:
    Partition Size Range of values in the offset column to use to create partitions.

    If the offset column is a Datetime column, provide the partition size in milliseconds. For example, to create a partition for every hour, enter 3600000.

    Available when using multithreaded partition processing.

    Max Partitions maximum number of partitions to be maintained or processed at one time for a single table. Adjusting this value can increase throughput depending on various factors, including the machine running Data Collector and the database server type and capacity.

    The minimum positive value is 2, to ensure the origin can make progress through the partitions.

    Enter -1 to use the default behavior, allowing the origin to create up to twice as many partitions for each table as threads used by the origin. Best practice is to start with the default behavior and adjust to tune performance.

    Available when using multithreaded partition processing.

    Offset Column Conditions Additional conditions that the origin uses to determine where to start reading data for this table configuration. The origin adds the defined condition to the WHERE clause of the SQL query.

    Use the expression language to define the conditions. For example, you can use the offset:column function to compare the value of an offset column.

  4. To enter JDBC credentials separately from the JDBC connection string, on the Credentials tab, configure the following properties:
    Credentials Property Description
    Username User name for the JDBC connection.
    Password Password for the JDBC account.
    Tip: To secure sensitive information such as user names and passwords, you can use runtime resources or credential stores.
  5. When using JDBC versions older than 4.0, on the Legacy Drivers tab, optionally configure the following properties:
    Legacy Drivers Property Description
    JDBC Class Driver Name Class name for the JDBC driver. Required for JDBC versions older than version 4.0.
    Connection Health Test Query Optional query to test the health of a connection. Recommended only when the JDBC version is older than 4.0.
  6. On the Advanced tab, optionally configure advanced properties.
    The defaults for these properties should work in most cases:
    Advanced Property Description
    Maximum Pool Size Maximum number of connections to create. Must be equal to or greater than the value of the Number of Threads property.

    Default is 1.

    Minimum Idle Connections Minimum number of connections to create and maintain. To define a fixed connection pool, set to the same value as Maximum Pool Size.

    Default is 1.

    Connection Timeout Maximum time to wait for a connection. Use a time constant in an expression to define the time increment.
    Default is 30 seconds, defined as follows:
    ${30 * SECONDS}
    Idle Timeout Maximum time to allow a connection to idle. Use a time constant in an expression to define the time increment.

    Use 0 to avoid removing any idle connections.

    When the entered value is close to or more than the maximum lifetime for a connection, Data Collector ignores the idle timeout.

    Default is 10 minutes, defined as follows:
    ${10 * MINUTES}
    Max Connection Lifetime Maximum lifetime for a connection. Use a time constant in an expression to define the time increment.

    Use 0 to set no maximum lifetime.

    When a maximum lifetime is set, the minimum valid value is 30 minutes.

    Default is 30 minutes, defined as follows:
    ${30 * MINUTES}
    Auto Commit Determines if auto-commit mode is enabled. In auto-commit mode, the database commits the data for each record.

    Default is disabled.

    Enforce Read-only Connection Creates read-only connections to avoid any type of write.

    Default is enabled. Disabling this property is not recommended.

    Transaction Isolation Transaction isolation level used to connect to the database.

    Default is the default transaction isolation level set for the database. You can override the database default by setting the level to any of the following:

    • Read committed
    • Read uncommitted
    • Repeatable read
    • Serializable
    Init Query SQL query to perform immediately after the stage connects to the database. Use to set up the database session as needed.
    Initial Table Order Strategy Initial order used to read the tables:
    • None - Reads the tables in the order that they are listed in the database.
    • Alphabetical - Reads the tables in alphabetical order.
    • Referential Constraints - Reads the tables based on the dependencies between the tables.
    On Unknown Type Action to take when the origin encounters a record with an unsupported data type:
    • Stop Pipeline - Stops the pipeline after completing processing the previous records.
    • Convert to String - Converts the data to string and continues processing.