Sciweavers

Share
VLDB
2000
ACM

ICICLES: Self-Tuning Samples for Approximate Query Answering

11 years 3 months ago
ICICLES: Self-Tuning Samples for Approximate Query Answering
Approximate query answering systems provide very fast alternatives to OLAP systems when applications are tolerant to small errors in query answers. Current sampling-based approaches to approximately answer aggregate queries over foreign key joins suffer from the following drawback. All tuples in relations are deemed equally important for answering queries even though, in reality, OLAP queries exhibit locality in their data access. Consequently, they may waste precious real estate by sampling tuples that are not required at all or required very rarely. In this paper, we introduce icicles, a new class of samples that tune themselves to a dynamic workload. Intuitively, the probability of a tuple being present in an icicle is proportional to its importance for answering queries in the workload. Therefore, an icicle consists of more tuples from a subset of the relation that is required to answer more queries in the workload. Consequently, the accuracy of approximate answers obtained by usi...
Venkatesh Ganti, Mong-Li Lee, Raghu Ramakrishnan
Added 26 Aug 2010
Updated 26 Aug 2010
Type Conference
Year 2000
Where VLDB
Authors Venkatesh Ganti, Mong-Li Lee, Raghu Ramakrishnan
Comments (0)
books