|
|
|
     |
                                     |
Return to Potpourri
, we proposed a tool called PLASTIC (Plan Selection Through Incremental Clustering) to significantly increase the scope of plan reuse. The tool is based on the observation that even queries which differ in projection, selection and join predicates, as also in the base tables themselves, may still have identical plan templates, that is, they share a common database operator tree. By identifying such similarities in the plan space, we can materially improve the utility of plan cacheing. In the demo, a Java implementation of PLASTIC working with Oracle 9i and IBM DB2 is presented, and its potential for amortizing query optimization overheads is highlighted.
@inproceedings {DBLP:conf/sigmod/SengarH03, author = {Vibhuti S. Sengar and Jayant R. Haritsa}, booktitle = {SIGMOD Conference}, title = {PLASTIC: Reducing Query Optimization Overheads through Plan Recycling.}, pages = {676}, year = {2003}, url = {db/conf/sigmod/sigmod2003.html#SengarH03}, ee = {http://www.acm.org/sigmod/sigmod03/eproceedings/papers/dem20.pdf}, crossref = {conf/sigmod/2003}, bibsource = {DBLP, http://dblp.uni-trier.de} }  ©2004 Association for Computing Machinery |
