Witness Node Quorum Service for Distributed Storage Consistency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Active/active storage systems face challenges in maintaining synchronization and data availability due to failures, particularly in geographically remote access nodes, where existing solutions like quorum disks are expensive and lack flexibility.
Innovation Solution
A distributed system with a witness node and membership service that uses majority voting to manage compute site membership changes during failures, ensuring continued joint computation by selecting viable subsets of compute sites, even in communication failures, and allowing rejoining of suspended sites.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If physical quorum disks are used to resolve tie-break situations in active/active storage systems, then reliability is improved, but device complexity and deployment cost increase
Solution Approach 1:
The patent replaces the physical quorum disk with a virtual copy implemented as software on ordinary storage devices. The quorum service is replicated across multiple nodes, and a virtual quorum disk is created through software coordination rather than requiring dedicated physical hardware. This maintains the tie-breaking functionality while eliminating the need for specialized physical quorum disks.
Solution Approach 2:
The patent substitutes the mechanical/physical quorum disk system with a software-based solution running on standard storage devices. Instead of relying on physical hardware to enforce quorum decisions, the system uses software protocols, network communication, and logical coordination mechanisms to achieve the same reliability objectives without the complexity of dedicated physical quorum infrastructure.
2Reliability
If traditional quorum disk solutions are deployed, then data availability is maintained during failures, but deployment cost and flexibility are reduced
Solution Approach 1:
The patent makes ordinary storage devices perform the additional function of providing quorum services. Instead of requiring specialized quorum disks, any storage device in the network can participate in quorum decision-making. This universal approach allows the same hardware to serve multiple purposes (data storage and quorum enforcement) thereby reducing deployment costs and increasing flexibility.
Solution Approach 2:
The patent implements a dynamic quorum system where the composition of quorum nodes can change over time. Nodes can be added or removed from the quorum group based on system requirements and failure conditions, rather than being fixed in advance. This dynamic configuration allows the system to adapt to changing conditions while maintaining data availability, providing greater deployment flexibility compared to static quorum disk solutions.
3Speed
If geographically remote access nodes are used to improve data access localization, then access speed is improved, but maintaining cache coherence and synchronization becomes more difficult
Solution Approach 1:
The patent introduces a quorum service as an intermediary mechanism that coordinates between geographically distributed access nodes. When remote nodes cache data locally to improve access speed, the quorum service acts as a mediator to ensure all nodes maintain consistent views of data state. This intermediary coordination layer manages the complexity of synchronization across remote locations while allowing each node to benefit from localized caching.
Data Source
AI summary
A system for managing distributed storage includes a collection of one or more compute sites that are connected and that perform a joint computation. A witness node is connected to the one or more compute sites. A membership service allows a change in collection membership, in response to a failure, to continue the joint computation without causing data inconsistency or corruption. If the membership service is unable to reach a decision on the change in collection membership in response to the failure, the witness node controls the decision on the change in collection membership to continue the joint computation. A distributed storage system provided by the system described herein may advantageously meet consistency guarantees and maximize data access even when individual compute sites and/or their interconnects fail.