Cloud Gateway Asynchronous Mirror for Data Resiliency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage systems face challenges in achieving high resiliency and cost-effectiveness, particularly when adding a third mirror for enhanced protection, as they often incur performance impacts and increased costs due to complex setups and geographical limitations.
Innovation Solution
A distributed data storage system employing a cloud-based asynchronous mirror, where data is synchronously mirrored between local sites and asynchronously replicated to a cloud storage system, using a combination of block-based and object-based protocols, allowing for geographic diversity and reduced costs by enabling disaster recovery almost anywhere.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a third mirror is added to enhance protection, then data resiliency is improved, but system complexity and cost increase
Solution Approach 1:
The patent introduces a cloud gateway as an intermediary component that mediates between the storage arrays and cloud storage systems. This gateway abstracts the complexity of cloud integration, providing a standardized interface for asynchronous replication while isolating the core storage system from cloud-specific complexities. The gateway handles protocol translation, data formatting, and cloud communication, thereby enabling third mirror functionality without proportionally increasing overall system complexity.
Solution Approach 2:
The patent employs asynchronous copying of data to cloud storage systems as the third mirror. Instead of implementing a full-featured storage array replica, the system creates simplified data copies in cloud storage. These cloud copies serve as disaster recovery mirrors with reduced complexity compared to traditional array-based mirrors, as they require fewer components and can be managed through automated replication processes rather than complex array coordination.
2Reliability
If synchronous mirroring is used between arrays, then data consistency is improved, but performance impact increases
Solution Approach 1:
The patent segments the replication functionality into two distinct modes: synchronous replication between local storage arrays for critical data consistency, and asynchronous replication to cloud storage for disaster recovery. This segmentation allows the system to apply synchronous protocols only where absolutely necessary (between arrays that both remain operational), while using less performance-intensive asynchronous protocols for cloud backup. The segmentation enables the system to maintain data consistency where needed without universally applying the performance overhead of synchronous protocols.
Solution Approach 2:
The patent applies partial synchronous action by implementing synchronous replication only between the two local arrays rather than involving the cloud storage in synchronous operations. The cloud storage serves as an asynchronous backup, performing exactly enough replication to ensure disaster recovery capability without the excessive performance cost of making it part of the synchronous replication chain. This partial application of synchronous protocols maintains consistency where critical while avoiding performance degradation in non-critical paths.
3Adaptability or versatility
If cloud backup is implemented, then geographic flexibility is improved, but network dependency increases
Solution Approach 1:
The patent implements preliminary local replication between the two storage arrays before relying on cloud backup. This ensures that even if network connectivity to the cloud is lost or interrupted, the local arrays maintain each other as backups and can continue operating. The cloud backup serves as an additional layer of geographic flexibility rather than the primary recovery mechanism, allowing the system to function independently of cloud network dependency while still benefiting from geographic distribution capabilities when available.
Data Source
AI summary
A distributed data storage system includes first and second data storage systems (DSSs) at respective sites and a cloud gateway at a third. The first DSS remotely mirrors data using synchronous and asynchronous channels both employing a block-based protocol (e.g., SCSI). The second DSS is connected to the synchronous channel to provide synchronous backup to the first DSS. The cloud gateway is connected to the asynchronous channel and has a cloud storage connection to a cloud storage system, the cloud storage connection using a wide area network interface and employing an object-based protocol. The cloud gateway (1) provides asynchronous backup to the first DSS via the asynchronous channel, and (2) uses the cloud storage connection to access object-based cloud data storage to store data of the asynchronous backup. The use of the cloud gateway and cloud storage system for an asynchronous third mirror enhances reliability/resiliency in a cost-effective manner.