Stephen Withers
Wednesday, 29 July 2009 13:52
IT Industry -
Development
Page 2 of 3
To give an indication of the relative performance, a certain query executed against a database with six million records took 16.5 seconds under Ingres 9. Switching to the VectorWise engine reduced this to just 0.2 seconds.
For comparison, a hand-written C++ program to do the same job took 0.04 seconds.
Jason Leonidas, managing director of Ingres ANZ told iTWire this development is "really game changing".
According to Leonidas, while it's too early to be sure, the new engine could be fast enough to allow data warehouse style work directly from a live database.
Traditional data warehousing is about building infrastructure around data to get the required performance. But VectorWise uses standard hardware and a standard database management system - "no proprietary infrastructure, all off-the-shelf hardware with a quantum leap in the way the processing is done," he said.
The technology eliminates the need for (slow) cross-table joins, indexing, or optimisation. Instead, it just "grabs the raw data... runs the process and it's done in seconds or minutes, not hours," Leonidas told iTWire.
What VectorWise will allow is a personal data mart, allowing senior executives and 'power' analysts to run complex, ad hoc queries without complex infrastructure and support requirements. While this is not an existing market, Ingres believes it is potentially a very large one.
Leonidas described a scenario where Excel can be used as the front-end to databases with millions of rows of data, returning results at "the speed of thought" (a phrase you'll probably be hearing a lot from Ingres people).
"It's going to be totally different to everything else that's out there," he told iTWire.
Latest Headlines from IT INDUSTRY
