- Abstract:
-
We present the case for holistic query evaluation: a novel query processing technique that is based on the generation of query-specific and hardware-friendly source code. The source code is compiled and dynamically linked to the database server for processing. By looking at how modern hardware works and identifying the performance bottlenecks, we have devised a collection of algorithm templates that we dynamically instantiate per individual query. Our algorithms cover a large portion of SQL: selections, projections, joins, and aggregation. On that foundation we build query optimization primitives. We present the design and development of a prototype system called HIQUE the Holistic Integreated Query Engine that incorporates our proposals. We undertake a detailed experimental study of the system s performance. The results show that the system s efficiency, in the majority of cases, surpasses both well-established and currently-emerging query processing alternatives. At the same time it does not affect other orthogonal aspects of a database system, such as the storage layer or concurrency control. The results, we believe, exhibit the potential of the holistic approach to optimize the data flow of query engines for main memory execution on contemporary hardware.
- Links To Paper
- 1st Link
- Bibtex format
- @Article{EDI-INF-RR-1295,
- author = {
Konstantinos Krikellas
and Stratis Viglas
and Marcelo Cintra
},
- title = {Holistic Query Evaluation: Optimizing the data flow of query engines for main memory execution},
- journal = {ACM TODS},
- year = 2008,
- url = {http://homepages.inf.ed.ac.uk/s0676486/index_files/HolisticQueryEvaluation.pdf},
- }
|
