Until now, data warehouses have concentrated mainly on analysing data from transaction-based systems, but as the market matures, users are increasingly keen to access other data sources and move away from technology-based solutions to pre-packaged applications.
Colin White, founder of consultants, Database Associates International, dubbed this trend, ?decision processing? (DP), at the Decision Processing 98 conference in Redwood City yesterday, and explained that, while it was aimed at generating information to help executives manage their business more effectively, it was causing IT problems of its own.
While DP is based on data warehouse technology, analytical applications and third party data sources are also seen as important elements of the mix. However, unless users integrate all of these different elements using a DP hub, White said they would end up with ?islands of information? that cannot be used on an enterprise-wide basis.
?Till now, people have focussed on data from transaction processing systems, but they now want to access data from their office systems, the Web and from third parties selling information to them," he explained.
"But, users won?t end up with all of this data in one data warehouse and as more analytical applications are introduced, you?ll see more data marts. This will create ?islands of information? and so you?ll need to integrate the pieces via technology such as an information portal,? he continued.
He added: ?A decision processing hub based on standard interfaces is necessary to bring all these pieces together and enables them to plug and play, whether you?re looking at a data warehouse, a database or an enterprise resource planning application. Users need to integrate these pieces or they?ll go out of business.?
To create a DP hub, White advised IT departments to use a common information model throughout their business and introduce standardised data capture interfaces and files as well as plug-compatible and interoperable data transformation routines across their systems. They should also introduce staging areas to hold shared information from different sources until it was needed.
To manage this information, users needed to introduce standard interfaces to all their data stores, create a hub for metadata interchange to take place and build common models for their metadata definitions.
Finally, to be able to analyse, model and distribute data across the business, users should implement plug-compatible and interoperable business intelligence tools and analytical applications, incorporate standardised data distribution files into their software and implement an information portal and business information directory at the front-end, so that users could find information easily.
The ghost is still in the machine
Campaigners want US authorities to break-up Instagram, WhatsApp and Messenger into separate companies
The perception of the industry as "a white man in a hard hat" is limiting new applicants, says Hayaatun Sillem
Almost two years late - and just as AMD is readying 7nm Zen 2 for early 2019