Mediator-less Distributed Storage Primary Bias Negotiation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-site distributed data storage systems face challenges in maintaining high availability when a mediator is temporarily unavailable, leading to connectivity issues and potential data consistency problems.
Innovation Solution
The system negotiates a primary bias state between primary and secondary storage sites, allowing them to operate independently without a mediator, ensuring non-disruptive operations by anchoring the primary I/O serving role at the configured source endpoint of a consistency group.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a mediator is used to manage failover between primary and secondary storage sites, then data consistency is maintained, but system availability deteriorates when the mediator becomes unavailable
Solution Approach 1:
The patent uses a mediator to coordinate failover decisions between primary and secondary storage sites, ensuring data consistency by preventing split-brain scenarios where both sites might simultaneously serve I/O operations. The mediator acts as a central authority that resolves conflicts and maintains consistent state across the distributed system.
Solution Approach 2:
The patent implements a mediator-less deployment mode where the primary storage site autonomously determines when to serve I/O operations without requiring mediator coordination. The system uses locally-available information (mediator availability status, I/O request timing) to make self-service decisions about when to accept or reject I/O requests, eliminating the single point of failure represented by the mediator.
2Productivity
If the system operates without a mediator (mediator-less deployment), then system availability is improved, but determining primary I/O serving authority becomes more complex
Solution Approach 1:
The patent establishes a pre-configured primary storage site in the mediator-less deployment mode before any failures occur. This primary site is predetermined based on configuration information stored locally at each storage site, eliminating the need for complex real-time negotiations about who should serve I/O operations when the mediator is unavailable.
Solution Approach 2:
The patent implements a feedback mechanism where the primary storage site monitors mediator availability status and adjusts its I/O serving behavior accordingly. When the mediator is available, the system operates in coordinated mode; when the mediator becomes unavailable, the primary site transitions to autonomous operation, using feedback about mediator status to determine when to accept or reject I/O requests.
3Productivity
If failover is allowed when mediator is unavailable, then system availability is maintained, but data consistency issues may occur
Solution Approach 1:
The patent prevents potential data consistency problems by having the primary storage site proactively reject I/O requests when the mediator is unavailable and it cannot confirm its authority to serve as the primary. This preliminary anti-action (rejecting I/O requests) prevents the harmful effect of split-brain scenarios where inconsistent data states could occur.
Solution Approach 2:
Instead of allowing failover by default and then managing conflicts, the patent inverts the approach by requiring explicit confirmation of primary status before allowing I/O operations in mediator-less mode. The system assumes a restrictive stance (reject I/O requests) and only permits operations when authority is explicitly established, reversing the traditional permissive failover approach.
Data Source
AI summary
Systems and methods include negotiating a primary bias state for primary and secondary storage sites for a mediator-less deployment. In one example, a computer-implemented method comprises detecting whether an active synchronous replication relationship is created for bi-directional synchronous replication between a first consistency group (CG1) of one or more storage members of the primary storage site and a second consistency group (CG2) of one or more storage members of the secondary storage site when a mediator is not configured for a mediator-less deployment or whether an existing mediator is being unconfigured with the mediator located remotely from the primary storage site and remotely from the secondary storage site, and negotiating a primary bias state and setting the primary bias state on a secondary storage cluster of the secondary storage site upon the active synchronous replication relationship being created with no configured mediator or when the existing mediator is unconfigured.