Object-Based Storage with Shared-Nothing Database
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data storage systems face inefficiencies in managing and searching large datasets, particularly in block I/O and file systems, where location changes require updates in access requests, and database management systems require consistent schemas for data retrieval, limiting scalability and performance.
Innovation Solution
Implementing an object-based storage system using a shared-nothing massively parallel processing database architecture, which enables parallel processing across multiple nodes, allowing for efficient data storage, retrieval, and search operations through object identifiers independent of physical or logical locations, and integrating with content addressable storage systems for enhanced performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If block I/O storage systems are used with location-based addressing, then data can be stored and retrieved using logical volume and block addresses, but the system requires updates to access requests when physical locations change, reducing efficiency
Solution Approach 1:
The patent introduces object identifiers as an intermediary layer between the host system and physical storage locations. Instead of directly addressing physical blocks, the system uses object identifiers that remain constant regardless of physical location changes. The storage system internally maps these identifiers to current physical locations, eliminating the need for host system updates when data moves.
Solution Approach 2:
The patent creates an abstract copy of the storage addressing model by using object identifiers that replicate the functionality of physical addresses without being tied to actual locations. This virtual addressing layer allows the system to manipulate location-independent references rather than physical addresses directly.
2Reliability
If database management systems with consistent schemas are used, then data retrieval can be performed systematically, but the systems require rigid schema structures that limit scalability and adaptability
Solution Approach 1:
The patent implements a dynamic data model where data structures can change without requiring schema modifications. The system allows data to be stored and retrieved with consistent reliability while adapting to changing data requirements through flexible, location-independent object identifiers that don't depend on rigid hierarchical schemas.
Solution Approach 2:
The patent segments the data access model into independent objects with unique identifiers, separating the identification function from the structural organization. This allows each data object to be accessed independently without requiring a comprehensive schema definition, enabling both consistency and flexibility simultaneously.
3Ease of operation
If traditional storage systems are used, then data can be organized in hierarchical file systems, but search operations become inefficient when dealing with large datasets distributed across multiple locations
Solution Approach 1:
The patent introduces object identifiers as an intermediary indexing mechanism that enables direct access to data objects without traversing hierarchical file structures. This intermediary layer provides both the organizational benefits of structured storage and the search efficiency of direct addressing through constant-time identifier-based lookup.
Data Source
AI summary
Methods, a system, and a program product for use in managing data storage is disclosed. In an embodiment, a method, system, and program product comprises receiving content via an object based storage interface to a data storage system; generating a database insert operation for an object derived from the content, the object having a generated object identifier; and using the database entry operation, storing data associated with the object in a shared-nothing database located within the data storage system, wherein the shared-nothing database is enabled to perform parallel processing across multiple independent nodes.In another embodiment, a method comprises receiving a delete command for an object via an object based interface to a data storage system; generating a database delete operation for the object based on the delete command; and using the database delete operation, deleting data associated with the object in a shared-nothing database located within the data storage system, wherein the shared-nothing database is enabled to perform parallel processing across multiple independent nodes.In yet another embodiment, a method comprises receiving a search command via an object based interface to a data storage system; generating a database query based on the search command; and using the database query, querying a shared-nothing database located within the data storage system, wherein the shared-nothing database is enabled to perform parallel processing across multiple independent nodes.