Distributed Hash Table Storage System for Data Reliability

The patent segments the storage system into multiple independent storage nodes forming a distributed cluster, eliminating the need for complex centralized RAID controllers. Each node operates independently with its own storage capacity, and data is distributed across nodes through hash-based addressing, reducing overall system complexity while maintaining reliability through redundancy.

The patent replaces the mechanical/controller-based RAID system with a software-based distributed hash table (DHT) architecture. Instead of relying on hardware controllers to manage redundancy and failover, the system uses cryptographic hash functions to deterministically route data to appropriate nodes, substituting mechanical control with algorithmic distribution.

The patent divides the storage system into multiple independent nodes that can be added or removed without affecting the entire system. Each node processes requests independently, distributing the processing load across the cluster. This segmentation allows storage capacity to scale independently from processing capability, as new nodes contribute both capacity and processing power.

The distributed hash table architecture enables storage nodes to autonomously route requests to the correct node based on hash calculations, eliminating the need for centralized controllers to manage data placement and retrieval. This self-service mechanism improves processing capability by distributing control and reducing bottlenecks.

The patent implements automatic fault detection and data recovery through the distributed hash table architecture. When a node fails, the system automatically detects the failure through missing hash responses and initiates data replication to alternative nodes without human intervention. This self-service recovery mechanism maintains data redundancy while eliminating manual maintenance time.

The system pre-determines data placement locations through hash calculations before failures occur. Each data block is associated with specific nodes based on hash values, so when failures happen, the system already knows which nodes should hold replica copies, enabling immediate recovery without complex rebuilding processes.

The patent creates a storage system where standard off-the-shelf hardware can be used for all nodes, eliminating the need for specialized FC SAN hardware. The distributed architecture provides universal functionality across different hardware platforms, allowing the system to achieve high performance through software optimization rather than expensive specialized components.

The system uses inexpensive standard storage nodes instead of expensive FC SAN hardware. Individual nodes can be added, removed, or replaced without affecting the entire system, allowing cost-effective scaling. The distributed architecture ensures that cheap individual components can be combined to create a high-performance system that rivals or exceeds expensive centralized solutions.