Fabric-Attached Storage Drives Distributed SAN Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage-area network (SAN) systems face performance bottlenecks due to centralized storage controllers, non-linear scalability, and increased latency in handling input/output requests, especially with large numbers of high-performance storage drives and client computing devices.
Innovation Solution
A distributed SAN system where fabric-attachable storage drives and client computing devices are directly connected to a fabric, with a SAN manager overseeing drive volume management and client software agents handling logical volume operations, allowing for decentralized storage functionality and reduced reliance on centralized controllers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If centralized storage controllers are used to manage storage drives, then storage management is simplified, but system scalability and performance deteriorate due to bottlenecks when handling large numbers of high-performance storage drives
Solution Approach 1:
The patent segments the centralized storage controller functionality into distributed components: fabric-attached storage drives operate autonomously, client computing devices execute SAN software agents, and a SAN manager handles high-level coordination. This segmentation eliminates the single-point bottleneck by distributing control across multiple independent nodes in the fabric.
Solution Approach 2:
The patent transitions from a hierarchical centralized control model to a peer-to-peer distributed model where storage drives, clients, and managers operate as equal participants on the fabric. This dimensional shift in control architecture allows parallel operations and eliminates the sequential processing bottleneck of centralized controllers.
2Reliability
If centralized storage controllers handle all I/O requests, then data consistency is maintained, but latency increases due to the single point of control
Solution Approach 1:
The patent introduces fabric-attached storage drives as intermediaries that directly handle data transfer between clients and storage media without requiring centralized controller intervention. The SAN software agent on the client and SAN manager provide coordination, but the actual data path bypasses the traditional centralized controller bottleneck.
Solution Approach 2:
Storage drives are equipped with autonomous capabilities to manage their own operations, including direct response to client I/O requests and self-coordination through the fabric. This self-service approach eliminates the need for every operation to traverse the centralized controller, reducing latency while maintaining consistency through fabric-based coordination protocols.
3Productivity
If high-performance storage controllers are deployed to handle increased I/O loads, then throughput improves temporarily, but system complexity and cost increase significantly
Solution Approach 1:
The patent extracts the heavy lifting of I/O processing from the centralized controller and distributes it to fabric-attached storage drives and client-based SAN software agents. This extraction eliminates the need for increasingly complex high-performance controllers by pushing processing capabilities to the edges of the storage system.
Solution Approach 2:
The fabric-attached storage drives and client devices serve multiple functions: storage drives provide both data storage and autonomous data transfer capabilities, while client devices function as both compute resources and storage access points through SAN software agents. This multi-functionality reduces the need for dedicated high-performance controller hardware.
4Productivity
If fabric-attached storage drives operate autonomously without centralized control, then scalability and performance improve, but coordination and data consistency management become more complex
Solution Approach 1:
The SAN manager implements feedback mechanisms to monitor and coordinate autonomous storage drives and clients on the fabric. This feedback loop enables the system to maintain data consistency and manage resources efficiently despite the distributed autonomous operation, allowing scalability without proportionally increasing coordination complexity.
Data Source
AI summary
A storage area network (SAN) includes fabric-attachable storage drives that are each directly connected to a fabric and that operate drive volumes. The SAN includes SAN software agents executed by client computing devices that are directly connected to the fabric. Each SAN software agent operates logical volumes realized by the drive volumes of the storage drives. The SAN includes a SAN manager directly connected to the fabric to manage the drive volumes of the SAN storage drives and to manage the logical volumes that the SAN software agents operate.