Storage Controller Architecture for Scalable Clustered Fabric
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage systems face challenges in scaling storage performance and capacity cost-effectively, as upgrading to a networked storage architecture requires expensive components and complex reconfiguration, making it prohibitive for smaller enterprises.
Innovation Solution
The implementation of a flexible storage controller architecture that enables scalable connectivity among multiple controllers through a switched fabric inter-controller connection, allowing for remote DMA access and cache coherency management, which facilitates dynamic reconfiguration from a single controller to a clustered environment without significant complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If storage systems upgrade to a networked storage architecture to scale storage performance and capacity, then storage capacity and performance are improved, but system cost and complexity increase significantly
Solution Approach 1:
The storage controller is designed with multi-functionality to operate in both standalone and networked modes. The same controller hardware and software can function as a single standalone unit or as part of a networked cluster, eliminating the need for separate controller types and reducing overall system complexity while maintaining scalability.
Solution Approach 2:
The patent implements a nested architecture where standalone storage controllers can be logically grouped and managed as part of a larger networked storage cluster. The cluster management software provides a unified layer that manages multiple controllers, allowing incremental scaling from single to multiple controllers without requiring complete architectural redesign.
2Reliability
If storage systems are configured with redundant storage controllers to enhance reliability and performance, then reliability and I/O processing capability are improved, but cache coherency management complexity and cost increase
Solution Approach 1:
Multiple storage controllers are merged into a logical cluster with unified management. The cluster management software combines the cache coherency management functions of individual controllers into a centralized coordination mechanism, allowing controllers to work together as a unified system while simplifying the management of cache coherence through shared metadata and coordinated operations.
Solution Approach 2:
The cluster management software acts as an intermediary layer between redundant storage controllers, handling cache coherency management, load balancing, and failover coordination. This intermediary abstracts the complexity of inter-controller communication and cache synchronization, allowing individual controllers to maintain their local caches while the intermediary ensures system-wide coherence.
3Reliability
If dedicated communication channels are provided between redundant storage controllers for cache coherency exchanges, then cache coherency management is improved, but system cost and complexity increase
Solution Approach 1:
The storage network fabric is designed to serve multiple functions simultaneously - it handles both data I/O operations and controller-to-controller communication for cache coherency. This multi-functional use of the existing network infrastructure eliminates the need for dedicated communication channels while maintaining reliable cache coherency management through the same fabric that handles storage traffic.
Data Source
AI summary
Systems and associated methods for flexible scalability of storage systems. In one aspect, a storage controller may include an interface to a fabric adapted to permit each storage controller coupled to the fabric to directly access memory mapped components of all other storage controllers coupled to the fabric. The CPU and other master device circuits within a storage controller may directly address memory an I/O devices directly coupled thereto within the same storage controller and may use RDMA features to directly address memory an I/O devices of other storage controllers through the fabric interface.