As Published In
Oracle Magazine
January/February 2008

AT ORACLE: Interview


Optimized for Data Warehousing

By Jeff Erickson

Oracle and partners provide preconfigured hardware and software for data warehousing.

Ray Roccaforte, Oracle vice president of database development for data warehousing, analytics, and business intelligence (BI), sat down with Oracle Magazine's Jeff Erickson to talk about data growth and the Oracle Optimized Warehouse. The following is an excerpt from that interview. Download a podcast of the full interview.

Oracle Magazine: Tell us about data growth and how it contributed to the Oracle Optimized Warehouse.

Roccaforte: Everybody's familiar with Moore's Law, which says that processor speeds are doubling every 18 months. But with those remarkable changes in processor speeds, why isn't our work productivity improving at a similar rate or at least at a more-rapid rate?

The answer is that data volumes are actually growing at a faster rate than Moore's Law. Ten years ago, it was unusual to see a data warehouse of 50 gigabytes. Today people are running Oracle data warehouses of hundreds of terabytes. So even though computers are getting much faster, computing is actually slowing down relative to the amount of data. This is going to have profound implications on how data is going to be managed and analyzed in the future.

One of the biggest implications is that with data volumes at such large levels, you really need to be able to scan data very rapidly, and that is what led to the Oracle Optimized Warehouse. In the Oracle Optimized Warehouse, we present the customer with an out-of-the-box, balanced configuration that is optimized for data warehousing. It's a complete data warehouse solution from Oracle and hardware partners including Dell Inc., IBM, and Sun. Oracle Database is preinstalled and preconfigured. The hardware is designed and tested to be balanced so that scan rates are very high and there's enough bandwidth to get the data to the processors at a fast enough rate to keep those processors busy.

Oracle Magazine: What kinds of customers is the Oracle Optimized Warehouse best suited for?

Roccaforte: There's a whole class of customers that we see coming on board here, as data warehousing has become much more popular. Today, everybody wants a data warehouse, because that's where you do your analytics.

But there's not as much expertise available compared to the number of data warehouses in the world, so there are a lot of emerging customers now who want something that's ready to use out-of-the-box. They don't want to design everything from scratch. And those are the customers that we're targeting with Oracle Optimized Warehouse.

Next Steps


 Listen to the podcast

 Learn more about the Oracle Optimized Warehouse Initiative

Oracle Magazine: You've stated that Oracle Database 11g, which is at the heart of the Oracle Optimized Warehouse Initiative, will fundamentally change BI. Tell us more about that. Roccaforte: One of the things that I'm very excited about is the integration in Oracle Database 11g between our online analytical processing (OLAP) technology and our materialized views technology, which is the classical approach to managing aggregated data in the Oracle data warehouse. This is profound because the basic operations of BI are dominated by aggregations.

Materialized views allow customers to precompute and store the aggregations that they're interested in. So at runtime they don't have to do the aggregation, and queries return orders of magnitude faster. But it's really difficult for customers to create all the materialized views they need. Even if they could create them, it would be very costly to keep them refreshed. The revolutionary step that we've taken is to integrate the materialized views technology with our OLAP cube technology. An Oracle OLAP 11g cube is a collection of thousands, or tens of thousands, of individual aggregations, all managed as a single object, stored compactly, and refreshed as a single object. By marrying these two technologies, we get the benefits of both. The precomputed aggregations in the cube give great query performance on a much larger set of queries, and the maintenance is handled by the Oracle database, which removes that burden from the customer.

 



Jeff Erickson is a senior editor with Oracle Publishing.


Send us your comments