Tree Awareness for Relational DBMS Kernels
| Autor: | Grust, Torsten, van Keulen, Maurice, Blanken, Henk, Grabs, Torsten, Schek, Hans-Jörg, Schenkel, Ralf, Weikum, Gerhard |
|---|---|
| Přispěvatelé: | Databases (Former) |
| Jazyk: | angličtina |
| Rok vydání: | 2003 |
| Předmět: |
Theoretical computer science
Memory hierarchy Computer science Relational database EWI-7668 InformationSystems_DATABASEMANAGEMENT Data structure computer.software_genre Data type Tree (data structure) Relational database management system Data mining Tuple ddc:004 DB-PRJPF: PATHFINDER DB-XMLDB: XML DATABASES Spatial analysis computer IR-63610 |
| Zdroj: | Intelligent Search on XML Data: Applications, Languages, Models, Implementations, and Benchmarks, 231-245 STARTPAGE=231;ENDPAGE=245;TITLE=Intelligent Search on XML Data Lecture Notes in Computer Science ISBN: 9783540407683 Intelligent Search on XML Data |
| ISSN: | 0302-9743 |
| DOI: | 10.1007/978-3-540-45194-5_16 |
| Popis: | Relational database management systems (RDBMSs) derive much of their efficiency from the versatility of their core data structure: tables of tuples. Such tables are simple enough to allow for an efficient representation on all levels of the memory hierarchy, yet sufficiently generic to host a wide range of data types. If one can devise mappings from a data type Tau to tables and from operations on Tau to relational queries, an RDBMS may be a premier implementation alternative. Temporal intervals, complex nested objects, and spatial data are sample instances for such types Tau . The key to efficiency of the relational approach is that the RDBMS is made aware of the specific properties of Tau . Typically, such awareness can be implemented in the form of index structures (e.g., R-trees [7] efficiently encode the inclusion and overlap of spatial objects) or query operators (e.g., the multi-predicate merge join [11] exploits knowledge about containment of nested intervals). This chapter applies this principle to the tree data type with the goal to turn RDBMSs into efficient XML and XPath processors [1]. The database system is supplied with a relational [8] XML document encoding, the XPath accelerator [5]. Encoded documents (1) are represented in relational tables, (2) can be efficiently indexed using index structures native to the RDBMS, namely B-trees, and (3) XPath queries may be mapped to SQL queries over these tables. The resulting purely relational XPath processor is efficient [5] and complete (supports all 13 XPath axes). We will show that an enhanced level of tree awareness, however, can lead to a query speed-up by an order of magnitude. Tree awareness is injected into the database kernel in terms of the staircase join operator, which is tuned to exploit the knowledge that the RDBMS operates over tables encoding treeshaped data. This is a local change to the database kernel: standard B-trees suffice to support the evaluation of staircase join and the query optimizer may treat staircase join much like other native join operators. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|