Ingres' Melbourne office overlooks the Melbourne Grand Prix circuit in Albert Park, so the company followed the lead of the the Grand Prix's 'ultimate speed comparison' handicap race between a C63 Mercedes, a V8 Supercar and a Formula 1 car.

While the 55 second headstart for the Mercedes and 24 seconds for the V8 saw all three vehicles cross the finishing line in short order, it turned out that the database shootout needed a bigger handicap.

An unspecified commercial database ("we can't actually say what it is because of the contract [which forbids the publication of comparative performance]," said John Ryan, Asia Pacific marketing manager at Ingres Australia) was given a 55 second handicap against Ingres VectorWise running the same query on the same data on identical hardware. VectorWise returned the result in less than one second, whereas 'brand X' took well over three minutes.

The test involved a 12GB database containing 72 million rows, based on the data from the DBT3 benchmark.

Jason Leonidas, vice president, Asia Pacific sales and services, explained that most DBMSes are generic and require a lot of tuning to suit any particular workload, whereas VectorWise is designed for excellent performance on ad hoc queries. It show that "what you should be using [in those situations] is a purpose-built database," ie VectorWise.

VectorWise promises rapid 'slicing and dicing' - see page 2.

---

It allows rapid 'slicing and dicing' without having to prepare cubes, indexes, aggregations or other concepts usually associated with business intelligence (BI), he told iTWire. Those tasks involve skilled staff and take time, and rely on knowing what questions will be asked.

But with VectorWise, "the power is in the ability to do this without all the infrastructure" and the advance planning.

Melbourne-based Yellowfin's BI software has been certified on VectorWise, and the products will be sold together as a complete enterprise reporting and analytics solution. The bottleneck with BI is usually the capacity to massage and explore the data, Yellowfin CEO and co-founder Glen Rabie told iTWire. "VectorWise compresses the delivery cycle," he said.

Since Yellowfin's product addresses the interface issues and allows users to build reports very quickly, "this combination [Yellowfin and VectorWise] is ideal for them," he said. "Most business people don't have a very clear idea of the questions they want to ask," but with Yellowfin and VectorWise they can ask questions as they arise without worrying if the preparation has been done. If the organisation's strategy changes, the relevant reports can be quickly obtained.

Mr Rabie said Yellowfin was able to deliver 192 different reports in just six weeks to a client that had budgeted for the job to take one year. Another instance saw the creation of four dashboards and a couple of dozen reports in three days. For comparison, the the 2010 TDWI (The Data Warehousing Institute) BI Benchmark Report found that it took an average of 6.6 weeks to build one complex report or dashboard.

Users have been conditioned by Google searches to expect quick responses to queries, and organisations are increasingly looking to replace canned reporting with user-driven ad hoc and on-demand reporting. Yellowfin and VectorWise make this possible without the need for a data warehouse. In particular, the Japanese market is "very excited" about what the combination of Yellowfin and VectorWise can do, claimed Mr Rabie. "The joint value proposition is so important."

Performance is known to be an important criterion when choosing a BI system. Mr Ryan pointed to BI Survey 8 (compiled by BI vendor Board) that showed it was the highest ranked criterion from a business-benefits perspective.

CONTINUED

---

Although a data warehouse is not required, it is still necessary to copy data from the production (transaction-oriented) database into a VectorWise database, but this can be done automatically and in near real-time, Mr Leonidas explained. It is not possible to get the performance required for both transaction processing and BI from one DBMS, he said.

Users of conventional BI products have found it can take as much as three days to build a cube, he claimed, and the typical IT approach of throwing additional hardware at a problem does not work in this situation: doubling the hardware does not halve processing time.

Like VectorWise, Yellowfin is "a clean, fresh architecture" that is lightweight and fast. "If the customer has the issue [the need for rapid ad hoc query processing], they will see the value [in the products]," he said.

Existing VectorWise uses include financial institutions in North America, Europe, and the Asia Pacific region, and they are all seeing performance improvements in the range of 50 to 100 times, said Mr Leonidas. "It's the biggest step change... that's ever occurred."

Ingres and Yellowfin's joint marketing site is here.

VectorWise recently set a new record (by a huge margin) for the TPC-H benchmark for 100GB of data.