NCR tears into the object relational arena

NCR Corp. has its sites set on a whole new object of desire. Sometime within the next two to five years, the company plans to incorporate object technology into the core Teradata relational data warehouse.

In doing this, NCR admits it is bucking the conventional wisdom that says there is no market for object-relational databases.

“NCR is identifying the business value for object-relational databases,” explained San Diego-based Randy Lea, assistant vice-president, Teradata marketing, in the computer systems group. “Our competitors release technology for technology’s sake and that doesn’t win on the market place. What wins is business solutions and revenue growth for the customer. We’re looking at using [objects] for business solutions like the identification of fingerprints and face recognition.”

NCR chairman and CEO Lars Nyberg used the NCR Partners User Conference held last month in San Diego to voice his confidence about the future need for this technology.

“We are absolutely convinced that within a couple of years, object relational technology within a database will be a real must.

“Soon audio and video objects will compliment the traditional data warehouse information. Your imagination is the only limiting factor. Video can be used to deliver entertainment along with information — infotainment. You just have to know where is the added value (in offering audio and video) and know how to distinguish yourself.”

Nyberg went on to suggest that storing large numbers of video images would be a natural use of Teradata, as the database is noted for its ability to scale. “We are the only show in town when it comes to really huge databases.”

The basis behind the Teradata object relational system (TDORS) will be technology purchased from the University of Wisconsin, and according to Lea it will be a good fit with the company’s existing offering.

“The University of Wisconsin’s technology is based on a shared-nothing architecture, which is the same concept behind Teradata, which does not share resources between nodes. Each Teradata node runs its own copy of Unix or NT.”

The object-relational support will be sold as an optional module, and will sit on top of the operating system and run roughly parallel to the core Teradata database and the Teradata application layer.

That architecture appeases the concerns of Anthony Bradley, program director for the Meta Group’s open computing strategic services. The Reston, Va.-based Bradley said he “would hate to see them corrupt the Teradata engine with object-relational functionality. The object-relational engine has to remain outside Teradata.”

Bradley disagrees with the viewpoint expressed by NCR representatives about the immediate future of object-relational databases. “I don’t believe at this point in time that within the next three to five years it will play a large role, ” he said. “Complex data in a data warehouse is not gong to be as important as the basic relational data that can be queried and aggregated.”

Merv Adrian, vice-president of information systems at the Giga Information Group in Santa Clara, Calif., agreed that object relational technology needs to carve out a niche.

“Object relational got here a little too soon. The business environment has to be moved. “

At the conference the future of object relational technology was exemplified by a debate between two Montreal-based Provigo staffers. Robert C. Mariano is in marketing, Jacques Guilbault is director, corporative data, technology and information systems.

“I can’t see the use of object relational technology right now. We need to see the business needs,” Guilbault said. “We have to go step by step, look at things on a day to day basis, but we also have to be the visionary group. Any kind of consumer-friendly technology is going to be exploited, but I don’t want to do [object relational databases].”

Mariano countered with: “From the marketing side, however, I can see the need for it. We might want to apply it to a mass accessibility medium so a core of people can react to it.

“We’re talking vision. There are always new ways to use old tools while we wait for the cranial chip to be invented,” he joked.

Waiting for the right time and the right invention is what NCR is doing, Lea said. “We could probably come out with technology sooner than we could come out with product, but when we announce the technology it will probably be the announcement of some core, key business solutions.”


NCR to offer parallel data mining

In additional to the object-relational module, upcoming products will increase the effectiveness of a complete Teradata solution, said Randy Lea, NCR assistant vice-president.

“We are coming out with a metadata product that adheres to XML with the Microsoft standard. We need to manage within our own internal products where the data is going, and interface with the various tool sets so we know where the data is coming from. We are releasing a product called Metadata Services this quarter to manage the metadata internally.

“An environment that we will be talking more about in the next six months will be Scalable Discovery. Scalable Discovery is new enhancements we are adding in the tool space and the database itself to do data mining much more efficiently and in parallel. Instead of just taking a look at data serially, in small sample sets, we are going to be looking at the entire sample set. Most of that technology is limited by the technology not being able to look at data all at once. That will probably be second half of next year.”