Stay in sync - Introducing Teradata Unity 13.10

Teradata Unity 13.10, the latest enabling technology of the Teradata Analytical Ecosystem was formally announced at PARTNERS Conference 2011.  The product’s focus is to simplify the analytical ecosystem by removing the everyday complexities involved in query management and data synchronization across multiple Teradata systems.  Teradata Unity delivers on this strategy by way of product automation.

Teradata Unity serves as an abstraction layer for users and applications making multiple Teradata systems appear as a single Teradata database instance.  Unity dynamically manages all the traditional activities around query routing and data synchronization creating a single integration layer for the Teradata ecosystem.

Core Components: 

Query Management –

Teradata Unity has two query management modes.  The first is Read Override mode.  Read Override is a session based, read-only routing mechanism based on defined routing rules.  This mode supports Named and Preferred routing.  Under the Preferred routing mode, in the event of a system failure Unity will automatically reroute queries to the available system. As such, Teradata Unity is a full replacement for the Teradata Query Director product.  No future enhancements will be applied to Teradata Query Director as Unity will now provide this functionality although Teradata Query Director will continue to be supported through Q2 of 2013.

The second routing mode in Teradata Unity is Read/Write mode.  As the name implies, this mode allows for both reads and writes to be applied through Teradata Unity.  In this mode, Unity will perform a light parsing of the statement to understand which SQL objects are being accessed.  Unity has a data dictionary identifying the location of SQL objects in the various Teradata systems.  Based on this information, the statements will be automatically routed to the appropriate system or systems. 

Consider the following example: An environment has three Teradata systems (A, B, and C) with Sales data existing on systems A and B while HR data exists only on system C.  A statement going through Unity accessing Sales data will be intelligently routed and applied to system A and B, not C.  In Read/Write mode, the users no longer need to worry about defining routing rules or concern themselves with which system they should be querying.  Unity will do this for the user automatically.

Both query management modes, Read Override and Read/Write, are configurable at the table level. 

Data Synchronization –

Teradata Unity provides data synchronization though a mechanism called SQL Multicast.  The SQL Multicast function will apply writes to all systems that are under Unity control keeping them all in sync.  This is not change data capture or a table copy!  This is an approach where SQL statements are applied to multiple target systems, based on location of the SQL objects.  With Unity, there is no concept of Primary/Secondary rather all systems are considered peers.  In the event of an outage, Unity will automatically route users and applications to the next available system.  Reads and writes will continue to go to the available system and all writes will be logged in the Unity Recovery Log.  Upon both systems becoming available, Unity will apply all the writes in the Recovery Log in the same order and operation as they were applied to the available system keeping all systems in sync.  The approach makes any outage completely transparent to the user or application.  Failover is handled automatically. 

Unity allows for data synchronization across multiple systems for both planned and unplanned events.  This creates the opportunity to perform rolling upgrades with no interruption to users or applications. 

Finally, Unity will synchronize any changes to a given table with that same table in other Teradata systems under Unity control.  In other words, if the Sales table exists in system A, B and C, a change to the sales table in System A will be automatically applied to Systems B and C as well.  This synchronization includes DDL, DML, and DCL.  

Auto Resubmit –

The Auto Resubmit feature of Teradata Unity continues to drive automation and remove complexity from the users.  In the event of an outage or database restart, Unity will automatically recognize where a query has failed and resubmit the remaining portions of the query to the next available system.  The Auto Resubmit feature essentially makes the outage or database restart transparent to the application or user. 

Below is a screenshot of the Unity user interface.  The picture shows the ability to manage each system under Unity control, check the status and take action.  The user can change the state of a given Teradata system by clicking on the following commands:  Freeze, Demote, Halt, or Deactivate.  Once one of those commands is issued, the user can then click the Restore function to bring the systems back in sync. 

The screenshot below shows the Teradata Unity dashboard.  The dashboard details system-level information of the systems under Unity control including version numbers and key configuration settings such as Recovery Log size, estimated allowable downtime and last checkpoint timestamp. The dashboard also displays performance information.  Figure X shows not only the Read/Write mix, but also the distribution of Reads amongst the systems under Unity control.

In addition to the core components described earlier, Teradata Unity 13.10 is fully integrated with Teradata Multi-System Manager (TMSM).  Integration with TMSM enables customers to view Unity system health as well as receive alerts, warnings and information regarding data loads, synchronization, and potential outages. 

For additional information on Unity 13.10, please go to Teradata.com and search for Unity.

Documentation/Collateral