Single Master Failover Protocol for Data Replication

The system segments data into multiple partitions that can be independently replicated across different machines. Each partition can have its own replica set, allowing granular control over data distribution and failure isolation. This segmentation enables the system to achieve high availability through replication while managing complexity by treating each partition as an independent unit.

The patent introduces a coordinator component that acts as an intermediary to manage the complexity of distributed data replication. The coordinator handles partition assignment, replica management, and failover coordination, centralizing control logic and shielding individual nodes from the overall system complexity. This intermediary approach allows the system to maintain high reliability while reducing the operational burden on individual components.

The system dynamically adjusts read routing based on consistency requirements. For strong consistency reads, the system routes requests to the master node or coordinates with replicas to ensure data freshness. For eventual consistency reads, the system can distribute reads across multiple replicas to improve throughput. This dynamic approach allows the system to optimize between productivity and measurement precision based on specific read operation requirements.

The patent implements different consistency parameters for different read operations. The system allows clients to specify consistency levels (e.g., strong consistency, eventual consistency) for each read operation, and adjusts the read routing and replica selection based on these parameters. This enables the system to achieve high throughput for non-critical reads while maintaining data precision for critical operations that require strong consistency.

The system segments write operations by partitioning the data space into multiple partitions, each with its own master node. This allows write operations to be distributed across multiple partition masters simultaneously, improving overall write throughput while maintaining write consistency within each partition. The segmentation approach enables parallel write processing without sacrificing the precision of write operations within individual partitions.

The patent introduces a partition dimension to the write operation model. Instead of having a single centralized master for all writes, the system creates multiple partition masters across different data dimensions (partitions). Write operations are routed to the appropriate partition master based on the data key, enabling parallel write processing across multiple nodes while maintaining consistency within each partition. This dimensional approach transforms the write scalability problem from a single-point bottleneck to a distributed parallel processing model.