Concurrent Read-Only Filesystem Access via Metadata Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data storage systems face challenges in improving performance and client capacity without incurring the expense of new hardware, as filesystem operations are complex and non-parallelizable, limiting the ability to enhance read-only access efficiently.
Innovation Solution
A storage server increases data service capacity by obtaining and processing metadata to locate and read data blocks from a remote storage subsystem, enabling concurrent read-only access and parallel processing without modifying the underlying filesystem.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If additional hardware is purchased to improve filesystem server performance, then processing speed and client capacity increase, but system cost increases
Solution Approach 1:
The patent segments the filesystem server functionality by separating read-only operations from read-write operations. Multiple read-only server instances are created that share the underlying filesystem, allowing parallel processing of read requests without requiring additional storage hardware. This segmentation enables horizontal scaling of read capacity while maintaining a single write authority.
Solution Approach 2:
The patent creates multiple copy instances of the read-only server that can simultaneously serve clients. These copies share the same underlying filesystem data but operate independently to handle read requests. The copying approach allows the system to increase service capacity for read operations without duplicating the expensive storage infrastructure.
2Productivity
If filesystem operations are parallelized to improve performance, then client capacity increases, but data consistency and filesystem integrity are compromised
Solution Approach 1:
The patent inverts the traditional master-slave relationship by making multiple read-only servers equal in status to the primary read-write server. All servers operate independently without hierarchical control, yet consistency is maintained through the immutable nature of read-only operations. This inversion eliminates the need for complex locking mechanisms while preserving data integrity.
Solution Approach 2:
The patent establishes read-only server instances before they need to serve clients, pre-configuring them with the necessary filesystem access permissions. By preparing multiple ready-to-serve instances in advance, the system can immediately handle increased client demand without dynamic provisioning complexity, maintaining consistency through pre-established access rules.
3Productivity
If read-only access is concentrated on a single server, then filesystem complexity remains low, but service capacity and response time are limited
Solution Approach 1:
The patent makes the read-only servers universal by enabling them to serve multiple purposes: they can handle read requests from multiple clients simultaneously, share the same underlying storage resources, and potentially migrate or replicate across different physical hosts. This multi-functionality increases service capacity without proportionally increasing architectural complexity.
Data Source
AI summary
A storage server obtains metadata to describe a filesystem, then processes the metadata to locate a data block and reads the data block from a remote storage subsystem. Apparatus and software implementing embodiments of the invention are also described and claimed.