Transcript Thanks to volunteer(ed?)s!

Thanks to volunteers!
Amit Gupta
Arvind Hulgeri
Aditya Phatak
Krishna Prasad
Jinesh Vora
Future Directions in Query
Optimization
Panel (the usual culprits):
Soumen Chakrabarti,
Anand Deshpande,
Krithi Ramamritham,
Sunita Sarawagi,
Shridhar Shukla,
S. Sudarshan (chair)
Current Scenario
Query optimization has come a long way
in the last two decades
Still an area of active research
Driving forces:
TPCD and friends -- bragging rights!
Query optimizers are still very expensive
Object relational DBS, Web, increasingly
complex DSS queries, Data mining
Future Directions - my
view
Cost of query optimization
exponential algorithms, scale very poorly
can we have good approximation algorithms
with guaranteed bounds?
Parallel and distributed databases
Search space is extremely large in general
How to partition data
How to partition operations
Future Directions (contd.)
System optimization for workloads
view and index maintenance
query result caching
Multi-query optimization
Scheduling issues
pipelining and MQO
dealing with concurrent large queries
Future Directions (contd.)
Semistructed data
Directories, XML, etc
Distributed query processing and the
web:
Dealing with failed sites, unpredictable delays
Alternative sources of data, site descriptions
Querying the web
WebSQL, WebOQL, .. (Mendelzon.., Shmueli..,
Laks..)
Future Directions (contd.)
Sources with query capabilities
eg. SAP on top of relational DB
SQL wrappers on top of legacy sources
Query optimization and data mining
beyond relational algebra?