Log-Based Data Storage for Content Management Systems

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional relational databases suffer from inefficiencies such as wasted space, time-consuming searches, and expensive maintenance due to their inherent design, particularly in content management systems (CMS) that rely on B-tree data structures and proprietary file formats.

Innovation Solution

Implementing a log-based data storage system that supports append-based write operations without disk seeks, enables O(1) time searches through an index, and uses platform-independent file formats like UNIX's tar format, along with a flexible node-based logical data tree structure for efficient data management.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If B-tree data structure is used for indexing in relational databases, then data can be organized and searched, but search operations require logarithmic time and write operations require many disk seek operations

Engineering Contradiction:
Improvesearch speedVSAvoidtime for write operations
Core Design Contradiction:
SpeedVSLoss of time

Solution Approach 1:

The patent segments the index into multiple index files that are sorted by randomly generated identifiers. This segmentation allows parallel processing and direct access to specific identifier ranges, eliminating the need for sequential logarithmic searches in a single B-tree structure.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Instead of organizing data by hierarchical B-tree nodes, the patent inverts the approach by using flat, sorted index files with randomly generated identifiers. This inversion eliminates the overhead of tree traversal and disk seeks associated with B-tree write operations.

Inventive Principle:
Principle #13The other way round (Inversion)

2Quantity of substance

If B-tree data structure is used for indexing, then data organization is achieved, but the structure wastes up to 30% space

Engineering Contradiction:
Improvestorage space utilizationVSAvoidindex structure complexity
Core Design Contradiction:
Quantity of substanceVSDevice complexity

Solution Approach 1:

The patent extracts the hierarchical B-tree structure and replaces it with flat index files. This extraction removes the inherent space waste of B-trees (empty nodes and pointers) while maintaining the essential function of organizing and searching data through sorted identifier ranges.

Inventive Principle:
Principle #2Taking out (Extraction)

3Reliability

If proprietary file formats are used in relational databases, then data storage is standardized, but backup, integration and maintenance become difficult and expensive

Engineering Contradiction:
Improvedata storage consistencyVSAvoidmaintenance ease
Core Design Contradiction:
ReliabilityVSEase of manufacture

Solution Approach 1:

The patent uses universal, platform-independent file formats (such as CSV, JSON, or SQL dump formats) that can be read by multiple tools and systems. This universality enables easy backup, integration, and maintenance without being locked into proprietary formats, while still maintaining data consistency through structured schemas.

Inventive Principle:
Principle #6Universality (Multi-functionality)

4Loss of time

If append-based write operations are implemented at the end of log files, then write operations require no disk seeks, but data structure searching is needed before writing

Engineering Contradiction:
Improvewrite operation timeVSAvoiddata structure complexity
Core Design Contradiction:
Loss of timeVSDevice complexity

Solution Approach 1:

The patent performs preliminary actions by pre-generating randomly distributed identifiers and pre-sorting index files by these identifiers before write operations. This preliminary organization eliminates the need for searching before appending, as new entries can be directly appended to the end of log files with their final positions predetermined by the sorted index structure.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2144177B1System and method for a log-based data storage
Publication Date: 2012.10.17 ADOBE SOFTWARE TRADING
  • EP2144177B1 patent drawingFigure 1A~1B
  • EP2144177B1 patent drawingFigure 2A
  • EP2144177B1 patent drawingFigure 2B

AI summary

Embodiments of the invention provide a high-performance content management system with a log-based data storage. To speed data operations, updates and deletes are appended to log-based data files. Seek operations are not needed as old data is not overwritten or removed at the time of write operations. Instead, the log-based data files are periodically optimized by processes that compact data entries and remove data files containing old data. In addition, an associated index with entries accessible in constant time is used to facilitate the process of locating data entries within the data files. Embodiments of the log-based data storage are implemented in a standard file format for ease of development and maintenance and can be compressed to further improve performance. In one embodiment, the log-based data storage corresponds to a node-based logical data tree that can be adapted to house a variety of data, including data from relational databases (Fig. 1).