Active-Active Storage Witness Arbitration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current witness technology in active-active data storage systems fails to account for the overall availability criteria of storage nodes when selecting a winner in case of communication loss, leading to suboptimal decision-making during failover scenarios.
Innovation Solution
Implement a preferred winner negotiation mechanism between nodes based on operational parameters such as configuration, state characteristics, and data replication status, with the witness arbitrating the winner based on the latest negotiation results during communication failures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If witness selection relies only on periodic state exchange messages and local health status, then the witness technology is simple to implement, but the decision-making during failover is suboptimal and does not account for overall node availability criteria
Solution Approach 1:
The system performs preliminary actions by having storage nodes continuously exchange capability information and negotiate preferred winner designations before any failure occurs. This pre-negotiation establishes a preferred winner based on comprehensive operational parameters, so that when a failure happens, the witness can make an informed decision without needing to evaluate all parameters in real-time, thus improving reliability while managing complexity.
Solution Approach 2:
The witness acts as an intermediary that receives and stores capability information from storage nodes, including data replication status, CPU capacity, memory capacity, and other operational parameters. The witness uses this pre-collected information to make authoritative failover decisions, mediating between multiple nodes with different capabilities and ensuring the most suitable node becomes the winner.
2Reliability
If the system considers multiple operational parameters for winner selection, then the failover decision is more robust and informed, but the negotiation and information exchange process becomes more complex
Solution Approach 1:
The system segments the complex evaluation process into two distinct phases: (1) a continuous pre-negotiation phase where nodes exchange capability information and establish preferred winner designations, and (2) a failover decision phase where the witness makes the actual selection. This segmentation allows comprehensive parameter evaluation to occur gradually over time rather than all at once during failure, reducing the complexity burden on the failover mechanism itself.
Solution Approach 2:
Comprehensive node evaluation is performed as a preliminary action through continuous capability information exchange and preferred winner negotiation between operational nodes. By the time a failure occurs, the witness already has pre-negotiated preferred winner designations based on multiple operational parameters, eliminating the need for complex real-time evaluation during the critical failover moment.
3Measurement precision
If nodes continuously exchange operational parameters to negotiate preferred winner, then the winner selection is more accurate, but the continuous communication consumes additional bandwidth and processing resources
Solution Approach 1:
The system implements periodic action by having storage nodes exchange capability information and negotiate preferred winner designations at regular intervals during normal operation. This periodic exchange maintains accurate winner selection data without requiring constant communication, balancing measurement precision with resource consumption by updating information periodically rather than continuously.
Data Source
AI summary
The system, devices, and methods disclosed herein relate to a dynamic, robust method for choosing a “winner” in an active-active data storage network. In the systems and methods disclosed herein, two or more intelligent nodes within an active-active data storage network periodically exchange operational parameters in an ongoing negotiation regarding who should be the winner in the event of a communication failure within the network. The winner is chosen dynamically based on the operational parameters. A witness is kept apprised of the winner. In the event of a communication failure between the two nodes, the winner is chosen by the witness based on the most recently negotiated lock file reported by one or both of the nodes.