User-Defined Function Execution on Database Tuples
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional user-defined function frameworks in databases invoke scalar functions once per tuple of input, leading to inefficiencies and performance concerns due to high overhead, especially when dealing with large volumes of data.
Innovation Solution
Implementing a mechanism to invoke user-defined scalar functions once for multiple tuples, allowing them to handle and compute results for multiple input tuples simultaneously, and optimizing the evaluation of these functions on run-length encoded data without decompression, as well as using analysis functions to eliminate unnecessary evaluations based on function properties.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If user-defined scalar functions are invoked once per tuple, then each tuple can be processed individually, but the overhead of function calls becomes excessively high when dealing with large volumes of data
Solution Approach 1:
The patent merges multiple function calls into a single invocation by enabling user-defined scalar functions to process multiple tuples simultaneously. The database system passes multiple input tuples to the function in one call, and the function returns multiple results, thereby reducing the total number of function calls from N (where N is the number of tuples) to a much smaller number of calls.
Solution Approach 2:
The patent implements multi-functionality by allowing a single user-defined scalar function to handle multiple tuples rather than requiring separate function calls for each tuple. This universal approach enables the function to process batches of data, making it more efficient for large datasets while maintaining the same functional behavior for individual tuples.
2Reliability
If functions are evaluated on all input data, then complete results are obtained, but unnecessary evaluations consume computational resources when function properties allow elimination
Solution Approach 1:
The patent applies preliminary action by using analysis functions to examine function properties and determine in advance whether a user-defined function needs to be evaluated for particular input data. The analysis function checks conditions such as whether the function should be eliminated based on its properties, and only evaluates the function when necessary, thereby avoiding unnecessary computational resources while ensuring complete results when evaluation is required.
Data Source
AI summary
A method for executing a user-defined function on a plurality of input database tuples. The method may include causing a processor to invoke the function once; and to compute results of the function for the plurality of database tuples.