Teradata

Note

We have a detailed howto for your first steps with SQL databases in DSS.

You might want to start with that Howto. The rest of this page is reference information for Teradata.

DSS supports the following features on Teradata:

  • Reading and writing datasets
  • Executing SQL recipes
  • Performing visual recipes in-database
  • Using in-database engine for charts

Please see below for limitations and detailed instructions

Installing the JDBC driver

The Teradata JDBC driver can be downloaded from Teradata website.

The driver is made of two JAR files:

  • terajdbc4.jar
  • tdgssconfig.jar

To install:

  • Copy both JAR files to the lib/jdbc subdirectory of the DSS data directory
  • Restart DSS

Connecting using LDAP authentication

By default, the Teradata connector uses the TD2 authentication mechanism. To use other kinds of authentication mechanism, enable the “Use custom JDBC URL”.

Enter “jdbc:teradata://YOUR-TERADATA-HOSTNAME/” as both the “connection URL” and “displayed URL”.

Add Advanced JDBC properties as documented in the Teradata JDBC driver documentation.

To login using LDAP, add the following properties:

  • LOGMECH = LDAP
  • LOGDATA = username@@password

Using per-user-credentials with LDAP authentication

First, get familiar with per-user-credentials: Connections security

Switch the connection to “per-user credentials mode”. Then add the following properties:

  • LOGMECH = LDAP
  • LOGDATA = %{pucUser}@@%{pucPassword}

The %{pucUser} and %{pucPassword} will be replaced by the per-user-credential login and password at runtime.

Limitations

In-database charts

Breakdown by “Quarter” and “Week” are not supported for in-database charts on Teradata. You can workaround by using the DSS charts (this will be slower)

Sort recipe

The Sort recipe is not supported on Teradata inputs. You can workaround by setting the engine of the recipe to DSS engine (this will be slower).

Note that sorting with a Teradata output will have no effect since Teradata does not preserve order on write.

Split recipe

The “Random dispatch of data” with “subset of columns” mode is not supported on Teradata. You can workaround by setting the engine of the recipe to DSS engine (this will be slower).

Parallel build of partitioned datasets

The first build job creating a partitioned dataset (either the first time, or subsequent times after the dataset has been cleared) must not be run on multiple partitions in parallel.

If this “first build”, which creates the table, is run on multiple partitions in parallel, some partitions may randomly fail. You can either set the parallelism of the recipe to 1, or first build a single partition before building others.

Fast sync using TDCH

Fast synchronization of datasets between Teradata and HDFS is possible using TDCH. Please see Teradata Connector For Hadoop