After a recent Exalead media event in Paris, Francois Bourdoncle and I were enjoying respective glasses of French wine and discussing search-based applications, a favorite topic. I asked him what aspect of the SBA vision was most compelling to him. His response was both interesting and expected. Expected because the answer was profound as is most of what Francois says. It was also interesting because of the implications. He said “the ability of search-based applications to decouple enterprise applications from usage was the most important aspect of SBAs.”So, what did he mean? Our challenge with accessing information today, as it has been for some decades, is that it resides across many different data management systems. Further, new data sources emerge constantly. The form of data can change as new elements are added to records, new content is added to web sites, new formats are imposed for office documents, etc. If we tightly bind our information access applications to these sources, such that the access system is intimately dependent upon the form of the underlying data, then accommodating new data or application/database migration projects incurs cascading costs and use disruptions. Further, we find that the life cycle of a core enterprise application is very long, multiple years, in fact, because of the expense and effort to deploy it. Yet, the lifecycle for information needs may be much different. Tactical activities may require access to certain information for only weeks or months. Or the need for information may be long term, but not aligned with the beginning or end use of an enterprise application. Hence, Francois’s vision for SBAs. The concept is to provide a flexible intermediate layer between information users and information sources that can accommodate conflicting requirements. Some capability examples that derive from the concept of decoupling:
• Easily and quickly accommodate new data sources without disrupting ongoing use of the system
• Support large scale user communities while accessing hundreds of terabytes of information
• Relate and group different bits of information where no pre-existing common identifier key for joining data exists
• Reducing information query response times while lowering the cost to process those queries
• Provide a data integration system that accommodates not only record data from enterprise applications but also content from the Web (both 1.0 and 2.0). So by using a search-based architectural approach, we believe we can accommodate a broader range of seemingly conflicting requirements than we could otherwise. It sets IT and the user community free of the information access constraints imposed by large scale, though rigid enterprise applications.