The Past, The Present, and The Future Dieter Gawlick Architect dieter.gawlick@oracle.com 05/04/2003 Lowell 2003
A Big Hole Users need a coherent model to present: The past - the history What was known/happened when? The present - the current state We are doing OK The future - know about ‘events’ ASAP We need to be on top of things! Applications: RTE, BAM, RFID, … 05/04/2003 Lowell 2003
What is The Problem? History - often 30 to 70% of coding, and still Incomplete Error prone No ‘separation of duty’ Future - event generation is a significant design, implementation, and maintenance challenge What should be published? Demand analysis Semantics Queuing, publish/subscribe is of limited help 05/04/2003 Lowell 2003
Recommendations Focus on theory Model must be general enough to handle SQL/XML/domain data Model needs to be complemented by General information distribution model (with auditing) Constraints, i.e., timely (re)action to ‘inconsistencies’ Security integrated with data, (re)actions, information distribution, and organizational structure Vocabularies, semantic context A lot of the engineering is understood, try to stay away There are still significant engineering challenges > 10**7 queries (subscription ) with complex conditions Distribution with many conflicting objectives Optimal, secure, and long term storage of/access to history 05/04/2003 Lowell 2003
Another Problem Why is it so hard to manage information? Models are incomplete (SQL, XML) Mapping to problems is often very complex – data, languages, transactions, security, visualization, …! Visualization: Try a domain that is challenging and fun, e.g., ISTD What are the fundamental flaws in ‘our’ approach? 05/04/2003 Lowell 2003
Questions Comments 05/04/2003 Lowell 2003