Dynamic Quorum Membership Changes in Distributed Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Distributed data storage systems face challenges in maintaining data consistency and availability due to the complexity and operational costs associated with managing multiple copies or replicas of data, which can lead to stale data issues and performance bottlenecks during membership changes in quorum-based models.
Innovation Solution
Implementing dynamic quorum membership changes in distributed storage systems, where a new quorum set is created by updating the protection group policy to include a new member, allowing access requests to be processed while ensuring data consistency through replication of previous updates, thereby minimizing network traffic and reducing the need for network-based durability checks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple copies or replicas of data are stored to increase availability, then data availability is improved, but the complexity and operational costs for maintaining consistent versions increase
Solution Approach 1:
The patent implements dynamic quorum membership changes where the quorum set can be modified at runtime by adding or removing members. This allows the system to adapt to changing conditions (such as node failures or availability requirements) without requiring complete reconfiguration of the replication system, thereby maintaining data availability while reducing the complexity of managing multiple replicas
Solution Approach 2:
The patent uses protection group policies that are established in advance to define quorum requirements and membership rules. By pre-configuring these policies, the system can quickly respond to failures or availability needs without performing complex real-time calculations, thus reducing operational complexity while maintaining high availability
2Reliability
If techniques are employed to slow or halt requests when a particular copy is not current to prevent stale data, then data consistency is improved, but the effectiveness of maintaining multiple copies for data availability is blunted
Solution Approach 1:
The patent implements a feedback mechanism where the system continuously monitors the current state of replicas and quorum membership. When a replica becomes stale or a member needs to be removed from the quorum, the system automatically adjusts the quorum set without blocking client requests. This feedback-driven approach maintains data consistency while preserving the effectiveness of multi-copy storage for availability
Solution Approach 2:
The quorum membership is dynamically adjusted based on the operational status of replicas. When a replica is marked as stale or unavailable, the system dynamically removes it from the quorum set, allowing reads to continue from remaining current replicas. This dynamic adjustment maintains consistency without blocking availability
3Reliability
If quorum-based models are used to ensure data consistency during replication, then data consistency is improved, but performance is slowed or blocked during membership changes
Solution Approach 1:
The patent pre-establishes protection group policies that define quorum requirements and member eligibility before changes occur. When a membership change is needed, the system can quickly reference these pre-defined policies rather than performing complex real-time validation, thereby maintaining data consistency while improving the speed of membership changes
Solution Approach 2:
The patent extracts the quorum consistency checking logic from the data path and handles it separately through protection group policies. This allows the main data operations to proceed at full speed while quorum consistency is maintained through background policy enforcement, reducing the performance impact during membership changes
Data Source
AI summary
A distributed database system may implement dynamic quorum group membership changes. In various embodiments, a quorum set may maintain a replica of a data object among group members according to a protection group policy for the data object. A group member may be identified as to be replaced. In response, a new quorum set may be created from the remaining group members and a new group member. The protection group policy may be updated to include the new group members such that subsequently received updates are maintained at both the previous to quorum set and the new quorum set. Previously received updates may be replicated on the new group member. Upon completion of replicating the previously received updates, the protection group policy for the data object may be revised such that subsequently received updates are maintained at the new quorum set.