Answer

How to select an MPP database: DB2 vs. Teradata

Should you run a massively parallel processing (MPP) database with DB2 or Teradata? Find out how to compare DB2 vs. Teradata and how to choose a database system and with advice from a data warehouse expert.

Mark Whitehorn
By
Published: 24 Jun 2009

IBM has its own massively parallel processing (MPP) database. How does it fare against Teradata's? Wouldn't it be prudent to go with that option for DB2 (relational) users?

Yes, DB2 is indeed available in  massively parallel processing (MPP) form and MPP is Teradata's natural operating mode. I wouldn't choose between them on the basis of manufacturer; I'd make the choice based on the requirements, because the two systems are designed to do different jobs.

DB2 is essentially designed to be a transactional database and comes in standard and MPP versions. Teradata is designed as a data warehousing system.

Unlike other warehouse systems, Teradata is based on the relational model and the use of MPP technology gives Teradata a huge edge in the real-time world.

With virtually every other warehouse system you load data into it by batch processing extract, transform and load (ETL) routines — typically overnight — and then there is a pause, often of several hours, while the data is indexed, aggregated and generally readied for querying. With Teradata's Enterprise Data Warehouse (EDW) you can load in data much more frequently and it is available for querying immediately.

So, if your goal is data warehousing with real-time analytics, check out Teradata. If you need a fast, MPP transactional database, check out DB2.

 

Dig Deeper on Database management