Stretched Volume N-Way Active-Active Storage Configuration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data storage systems face challenges in providing seamless and efficient data access and replication across multiple host processors, as they often require direct addressing of physical storage devices, which can lead to complexity and inefficiency in managing shared data storage resources.
Innovation Solution
The technique involves configuring a 'stretched volume' across multiple data storage systems, where writes are synchronously replicated across all systems, ensuring data consistency and availability by using a hierarchical ranking system and lock management to coordinate access and ownership of logical addresses, allowing multiple hosts to access data as a unified logical volume.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple data storage systems are connected to provide common data storage for host processors, then data availability and redundancy are improved, but system complexity and difficulty in managing shared resources increase
Solution Approach 1:
The patent combines N volumes from N different data storage systems into a single stretched volume that appears as one unified logical device to hosts. This merging approach maintains data availability across multiple systems while simplifying the interface to hosts, as they see a single volume rather than managing multiple separate storage systems.
Solution Approach 2:
The patent introduces a stretched volume as an intermediary layer between hosts and the underlying N volumes across multiple storage systems. This intermediary abstracts the complexity of distributed storage management from hosts, while enabling data to be accessed and replicated across N systems through coordinated write operations and lock management.
2Speed
If direct addressing of physical storage devices is used by host processors, then data access efficiency is improved, but complexity in managing shared data storage resources increases
Solution Approach 1:
The stretched volume provides a universal interface that allows hosts to access data through a single logical volume regardless of the underlying distributed storage architecture. This multi-functional approach enables both efficient data access and simplified resource management by presenting a unified view of storage resources across N systems.
3Stability of the object's composition
If synchronous replication is implemented across multiple data storage systems, then data consistency is improved, but coordination overhead and processing time increase
Solution Approach 1:
The patent implements preliminary lock acquisition before write operations on the stretched volume. By acquiring locks from all N data storage systems in advance and establishing a hierarchical ranking system, the system prepares the coordination structure beforehand, reducing the actual coordination overhead during data replication operations.
Solution Approach 2:
The patent introduces asymmetric hierarchical ranking among N data storage systems, where each system has a unique rank. This asymmetry enables efficient lock management and write coordination by providing a deterministic order for acquiring and releasing locks, reducing the complexity of synchronous replication coordination across multiple systems.
Data Source
AI summary
A stretched volume may be configured from N volumes of N data storage systems configured as a cluster. N may be an integer greater than two. Each of the N volumes may be included in a different one of the N data storage systems. The N volumes may be exposed to a host as a logical volume having a unique identifier over a plurality of paths from the N data storage systems. The N volumes may be configured for multidirectional synchronous replication. At a first system of the cluster, a write operation may be received that writes to a target logical address of the stretched volume. Servicing the first write may include synchronously replicating the first write to every other one of the N data storage systems of the cluster. Also described are techniques for handling lock contention and avoiding deadlock in connection with processing writes to the stretched volume.