Database Page Grouping for Sequential I/O Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Database systems face inefficiencies in writing modified data from a cache to a data volume, leading to difficulties in later reading the data due to random scattering of data pages across volumes, resulting in a large number of disk seeks.
Innovation Solution
The system identifies pages to be written, creates sorted page groups by file and key, and writes them to available disk clusters, while monitoring and managing data volumes to optimize storage by grouping and updating available clusters, ensuring efficient writing and reading.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If modified data is written sequentially to the first free storage blocks available in the data volumes, then writing efficiency is improved, but reading efficiency deteriorates due to random scattering of data pages
Solution Approach 1:
The patent segments data pages into page groups based on their file identifiers and key ranges. Each page group contains pages that can be written contiguously to the data volume, creating organized segments rather than random scattering. This segmentation allows efficient sequential writing while maintaining readability through structured organization.
Solution Approach 2:
The patent performs preliminary sorting of pages into page groups before writing to the data volume. By pre-organizing pages according to file identifiers and key sequences, the system prepares data in an optimal format for both efficient writing and subsequent reading, eliminating the need for random disk seeks during read operations.
2Ease of manufacture
If data pages are randomly scattered over the data volumes for storage, then writing simplified data is improved, but reading complexity increases due to required disk seeks
Solution Approach 1:
The patent divides data pages into structured segments (page groups) organized by file identifiers and key ranges. This segmentation creates a systematic storage pattern that simplifies the writing process while enabling efficient reading through predictable data locations, reducing the need for complex disk seek operations.
Solution Approach 2:
The patent changes the organizational parameters of data storage by grouping pages according to file identifiers and key sequences rather than using random placement. This parameter change transforms the storage structure from scattered to organized, simplifying both writing operations and reading operations by eliminating random disk seeks.
3Reliability
If a large number of disk seeks are performed to retrieve data from different pages, then data retrieval completeness is improved, but reading efficiency deteriorates
Solution Approach 1:
The patent performs preliminary organization of data pages into sequential page groups before storage. This pre-organization ensures that related data pages are stored contiguously, allowing complete data retrieval through sequential access rather than multiple random disk seeks, thereby maintaining retrieval completeness while dramatically improving reading efficiency.
Solution Approach 2:
The patent merges related data pages into page groups based on file identifiers and key ranges. By combining pages that belong together into contiguous storage units, the system enables complete data retrieval through fewer, more efficient sequential operations rather than numerous separate random access operations.
Data Source
AI summary
A system, method, apparatus, means and computer program code for writing data includes identifying a plurality of pages to be written to a data volume, each page including a file identifier and a key, creating a first page group from the plurality of pages, the first page group including pages associated with a first file as identified by the file identifier and sorted by the key; and writing the first page group to the data volume.