ROWID Elimination Rewrite for Offload Engine Memory Pressure
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern RDBMS query optimizers introduce ROWIDs for query unnesting, which are not efficiently processed by offload engines due to memory pressure and hardware acceleration limitations, making query execution infeasible.
Innovation Solution
The ROWID Elimination Rewrite technique substitutes or removes ROWIDs with alternative columns, such as primary keys, to reduce memory pressure and enable query execution in offload engines, specifically applicable to TPC-H Benchmark queries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If query unnesting rewrite techniques are applied using ROWIDs, then query execution speed is improved, but memory pressure increases and offload engine compatibility deteriorates
Solution Approach 1:
The patent changes the data type parameter from ROWID (10 bytes) to alternative column identifiers (typically 4-8 bytes), reducing memory footprint while maintaining the query unnesting functionality. This parameter change enables offload engine compatibility without sacrificing execution speed improvements.
Solution Approach 2:
The patent replaces the expensive ROWID data structure with cheaper alternative column identifiers that consume less memory. These alternative identifiers serve the same functional purpose but with reduced resource requirements, making them suitable for offload engine environments with constrained memory resources.
2Productivity
If ROWIDs are introduced for query unnesting, then query optimization is improved, but memory footprint increases
Solution Approach 1:
The patent changes the size parameter of the identifier from 10 bytes (ROWID) to smaller alternative identifiers (4-8 bytes), directly reducing memory footprint while preserving query optimization capabilities through alternative unnesting techniques.
Solution Approach 2:
The patent uses alternative column identifiers that copy the essential identification functionality of ROWIDs without duplicating their memory-intensive structure. The alternative identifiers serve as functional copies that achieve the same query unnesting purpose with reduced memory consumption.
3Productivity
If ROWID pseudo-columns are used in rewritten queries, then query execution plan quality is improved, but hardware acceleration support deteriorates
Solution Approach 1:
The patent replaces ROWID pseudo-columns with simpler alternative column identifiers that are more amenable to hardware acceleration. The alternative identifiers have simpler data structures that can be more efficiently processed by hardware-accelerated offload engines, maintaining execution plan quality while enabling hardware support.
Solution Approach 2:
The patent changes the data structure parameter from complex ROWID format to simpler alternative identifier formats, making the data more suitable for hardware acceleration. This parameter change enables offload engines to apply hardware-accelerated operations while preserving query execution plan quality.
Data Source
AI summary
Techniques described herein propose a ROWID Elimination Rewrite that uses functional dependencies to substitute and/or eliminate ROWID pseudo-columns referenced in a query in order to reduce memory pressure and speed up processing.