|
Group D: New Environments and
Data Management System Issues
Mission Statement
Determine how information systems can exploit and operate in a computing environment that is increasingly interconnected, distributed, heterogeneous and dynamic.
Big ‘n' Wide
Large number of entities, wide area
many rules, distributed events
100s of DBs interoperating at a company
#'s of clients - bring your own cycles
middleware
workflows across autonomous systems
wide area consistency management
Smart Shopper DB
Cost and Quality conscious
Tradeoffs on latency, concurrency, correctness, completeness, resource usage (including user time).
The database that never forgets
Personal/group archive:
Data live-ness: media rollover
Multi-schema support
Locating data
No-fuss data management
Rich set of choices for physical organization and access methods; have to spoon-feed databases; have to manage data after extraction
Automated tuning
Reorganization tools
System configuration + reconfiguration
Easy in and out
Reduce cost ($+m) of database ownership
Cradle-to-grave data management
Really conception-to-grave, data is never "outside" the database
Direct capture instead of store+load
Necessary to do provenance
OS knows about all processing, why not DB knows about all data
Data logistics
Tend to view DB as a static thing, but the value of data is only realized when it moves
Data product manufacturing
Adaptive dissemination
Value-added brokering, reselling, pressing
Zero latency, instant data
Variable infrastructure
Push and broadcast
All the data all the time
Can reach every piece of data from every place
Never-fail
Connectivity
Media conversion
Spare Slogans
Data addiction/data mainlining
Knowledge Systems DB
We put you in the driver's seat: interactive query formulation and answering.
Application-aware information management
Database takes responsibility for application characteristics
Knows about end-to-end performance
Knows about quality requirements and can negotiate trade-offs
Fast wrong answers
Event detection
Aware of user interface
Improves application characteristics – recovery
Application models (equivalent of schemas) that lead to automated management
Applications
International criminal DB
All the data all the time
Telecommunications
Zero latency
Workflows across autonomous systems
wide are consistency management
Medicine, Digital Patient
Immediate collection of trauma, emergency data, more patient records, data capture, data staging
Virtual enterprise, personal department store
Information commerce
Digital globe
Producer side: Earth representation, 1m, 1pixel = 1 byte, ¾ ocean, fixed 10PB and EOS generates 4TB a day
Consumer side: 40M kids, 100 images, 800PB delivered each day
Digital city presented at the PI meeting.
Modes of Research
DB/Medical informatics collaborations
Intra-CS collaborations
OS, Distributed Systems, Networks, Languages, Software Architecture
Counteracting conservatism among reviewers
Speculative studies in context of initiative
Extracting industry experience
Developer/Academic workshops | 
