1
play

1 Language processing Query Graph Model Two stages: compilation - PDF document

Outline Extensible Query Processing in Motivation Starburst Solution: Extensible DBMS Language Processing Query Graph Model Presentation: Kati Query rewrite Discussion: Andrew Summary Motivation Starburst Project


  1. Outline Extensible Query Processing in � Motivation Starburst � Solution: Extensible DBMS � Language Processing � Query Graph Model Presentation: Kati � Query rewrite Discussion: Andrew � Summary Motivation Starburst Project � DBMSs disability to support other � Extensibility applications than administrative ones � Language extensions � No sufficient support for the � Internal processing extensions functions and data types needed by � Data management extensions the engineering’s applications � Worth noting the time when it came up � additional functions and data � Object-oriented phase types needed Two major components Starburst’s Language: Hydrogen � Corona: the query language � Based on SQL processor � Orthogonal � Core: data manager � Extensible � Table expression � Table function � very complex queries possible 1

  2. Language processing Query Graph Model � Two stages: compilation and • Vertices execution • Edges • boxes Query Rewrite Rule-based Approach � A form of optimization and a big � Rule language is C challenge � Two parts: condition and action, � New tranformations required each written a C function � Rule-based approach � Consistency � Creation of new rule system � Rule classes � Modularization � Greater scope of optimization and improved execution plan Rules- three classes Rule engine � Predicate migration � Independent of any rules � Projection push-down � Forward chaining � Operation merging � Several control strategies: � Sequential � Priority � Statistical � budget 2

  3. No! cost-based query-rewrite Cost-based optimization � All alternatives are generated � Plan generation � At the plan level cost-based � Plan costing � BUT interaction desired � Search strategy � SINCE number of alternatives � Designed to be orthogonal grows tremendously Plan generator Summary � Starburst: extensible DBMS � Strategy alternative rules (STARs) � Extensions to the language, the � A general-purpose STAR evaluator language processing and the data � A search strategy that chooses the next manager STAR to evaluate � Tabel expressions allow � An array of STARs orthogonality � Orthogonality & Extensibility � complex queries possible Summary � Query internally a QGM � QGM simplifies the DBC’s task, give him a great deal of flexibility and power � Rule-based query rewrite � Grammar-like rules to generate plans 3

Recommend


More recommend