Deferred XPath Evaluation in Binary XML Query Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Optimizing queries on views that provide abstraction over XML documents in database systems is challenging due to the potential adverse effects of transformation order on query execution performance, particularly when XML re-writing occurs before view merge, leading to unnecessary expression evaluation and reduced optimization opportunities.
Innovation Solution
The query optimizer determines the relative order of query transformations, considering the availability of a XML index to defer XML re-writing until after the view merge stage, ensuring that only necessary XPath expressions are evaluated and optimized, thereby improving query execution efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If XML re-writing is performed before view merge, then XPath expressions are transformed early, but unnecessary expressions are evaluated and optimization opportunities are reduced
Solution Approach 1:
The query optimizer performs preliminary analysis of the query plan to identify which XPath expressions are actually needed before executing the view merge. By determining the necessary expressions in advance, the system avoids unnecessary XML re-writing operations and computational waste while maintaining optimization opportunities.
Solution Approach 2:
The system dynamically adjusts the order of query transformations based on the specific query structure and available XML indexes. The query optimizer flexibly switches between different transformation sequences (XML re-writing before or after view merge) depending on what produces the most efficient execution plan, rather than following a fixed order.
2Productivity
If XML re-writing is performed before view merge, then transformations are applied early, but view merge optimization opportunities are reduced
Solution Approach 1:
The query optimizer performs preliminary analysis to determine which XPath expressions are needed before executing view merge. This preliminary determination allows the system to maintain full optimization opportunities while avoiding unnecessary transformations, effectively managing complexity by being selective about when transformations are applied.
Solution Approach 2:
The system changes the parameter of transformation timing based on query characteristics. By adjusting whether XML re-writing occurs before or after view merge, and by controlling which expressions are rewritten, the system optimizes the balance between transformation benefits and optimization opportunities.
3Reliability
If all XPath expressions are evaluated, then complete data is processed, but unnecessary expressions consume resources
Solution Approach 1:
The query optimizer extracts and identifies only the necessary XPath expressions that are actually needed for the query result. By taking out the unnecessary expressions from the evaluation process, the system maintains data completeness for required fields while eliminating resource consumption on unnecessary computations.
Solution Approach 2:
Instead of evaluating all XPath expressions completely, the system performs partial evaluation only on the subset of expressions that are actually needed. This partial action approach ensures sufficient data processing for the query requirements while avoiding the excessive energy consumption that would result from evaluating all expressions.
Data Source
AI summary
A mechanism is provided for improving the performance of particular database queries operating on a view comprising binary-encoded XML. A determination is made about whether a XML index is available to improve the computation of XPath data before deriving an optimized execution plan for a particular query. When a XML index is not available to help expedite the computation of binary-encoded XML, then deferring the computation of the XPath data until after the view merge stage avoids unnecessary computation of XPaths that are part of the view but not needed for the particular query.