ScoutMaster uses Caché to create a professional research engine
Key Benefits
- High Performance
- Cost-Effectiveness
- Platform Portability
- Web Connectivity
- Multidimensional Data Model
Most search engines function with all of the accuracy of a shotgun. They return several thousand hits even when a user enters a precise query. All the user can do is sift through each hit to determine if a document is even marginally related to the actual search term. Often it is impossible to determine why a page was returned at all.
Scout Research Systems GmbH has solved this problem neatly, using Caché's post-relational technology for its ScoutMaster professional research engine. ScoutMaster was designed specifically for people doing professional research on the Internet. Caché is the e-DBMS, a post-relational database system from InterSystems optimized for the unique requirements the Web places on applications like ScoutMaster.
Alfred Fraas, Scout Research's head of development, said that Caché was chosen for the underlying technology of the Web application primarily because of the database's multidimensional data model. "The multidimensional data model is one of Caché's special features that not only sets it apart from "ordinary" relational models but also from the purely object-oriented ones," Fraas explained. "This model simplifies the difficult process of data modeling by depicting data in multidimensional structures instead of conventional "flat" relational tables."
"The multidimensional data model is one of Caché's special features that not only sets it apart from "ordinary" relational models but also from the purely object-oriented ones."
- Alfred Fraas
Head of Development
Caché's multidimensional data model stores data in a manner that is similar to its appearance in the real world, which eliminates the need to compile various data from many tables in a time-consuming JOIN procedure. Instead, Caché's multidimensional data model stores data in a collection of sparse arrays with any number of subscripts. While one subscript could be an integer, another could have a meaningful name. This model is extremely useful because ScoutMaster must correlate multiple criteria that users enter. "This makes it possible to gain a great deal of speed, particularly in complex search operations," Fraas said
While Caché's multidimensional data model and high performance were the chief reasons ScoutMaster was built on Caché, Fraas was also impressed with Caché's Web connectivity that allows for easy session management over the Internet.
Caché's platform portability plays a key role in ScoutMaster's configuration. ScoutMaster distributes the workload across multiple systems, running Windows NT on the front end and Linux on the back end. Distributed Cache Protocol and Dynamic Namespace Mapping, two key technologies embedded in Cache, enables either end to be dynamically expanded and configured.
And finally, as Fraas explains it, Caché's "economical interaction" with hardware makes it a very attractive solution for Web applications. "It's possible to run ScoutMaster on a standalone NT server as well as on a UNIX cluster," he said.
"It is impossible to bring order into the World Wide Web," said Fraas. "One must accept the prevailing chaos on the Internet and search using the right search strategies. It is then possible to find all the information needed. We have come a great deal closer toward achieving this goal with the development of ScoutMaster with the Caché database."