90 likes | 234 Views
Relation Caching. Ravindra N. Guravannavar. Introduction. Relation Caching is a logical caching technique Reduces the client-server data transfer by exploiting the client resources Can be employed in a query-shipping architecture. Applicability. Select-Project queries on a single relation
E N D
Relation Caching Ravindra N. Guravannavar
Introduction • Relation Caching is a logical caching technique • Reduces the client-server data transfer by exploiting the client resources • Can be employed in a query-shipping architecture
Applicability • Select-Project queries on a single relation • Select condition is conjunctive • Predicates are simple • Large portion of the total number of attributes are projected
Example Application Data merchandizing over the Web R1(Cname, Turnover, EmpStrength, YearStarted, Business) Select * from R1 where YearStarted<=1995 AND Turnover>=2000000 AND EmpStrength>=10000
The key idea • Keep semantic description of the cache contents • Exploit the semantics expressed in the associative queries • Fetch only the remainder query result Sal Probe Remainder Q2 8000 5000 Q1 25 35 Age
Replacement policies Semantic distance function -5 0 -3 -2 0 0 After Q1 After Q2 After Q3
Forming the remainder query R-Tree index on the semantic regions • Efficient searching of spatial data • Internal nodes have n-dimensional guiding rectangles • Each leaf node entry has the bounding box of a semantic region and a pointer to the set of tuples falling in that region
Implementation • CacheRelation manager • Transparent relation scans • Cache management • Query evaluation • Parser and Catalog manager
Scope for future work • Fixed size semantic regions for efficient cache utilization • Relaxing the constraints on the queries • Supporting JDBC interface