Peer-to-Peer Backup Storage with Distributed Hash Table Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data restoration processes in computer systems are inefficient due to network latencies, high costs, and the unavailability of mobile hosts, especially when using tape-based backups and centralized backup servers, which also fail to account for varying host capabilities and network connections.
Innovation Solution
A distributed system where hosts are organized into peer groups using a distributed hash table to store and manage data segments, allowing hosts to share backup responsibilities, negotiate storage capacity, and select peers based on availability and connectivity for efficient data distribution and redundancy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If centralized backup servers are used to store backup data, then data security and management control are improved, but network latency increases and restoration efficiency deteriorates
Solution Approach 1:
The patent segments backup data into multiple data segments and distributes them across multiple peer hosts in a peer-to-peer network. Each host stores portions of the backup data locally, eliminating the single point of centralized storage. This segmentation allows restoration to proceed in parallel from multiple sources, significantly reducing network latency and restoration time while maintaining data security through distributed storage and fingerprint-based verification.
2Quantity of substance
If tape-based backup media are used to minimize storage costs, then storage expense is reduced, but restoration speed and accessibility deteriorate due to manual mounting requirements
Solution Approach 1:
The patent replaces the mechanical tape-based backup system with a digital peer-to-peer network storage system. Instead of physically mounting tapes to access data, the system uses networked hosts with local storage to hold backup data segments. This substitution eliminates manual intervention for data access, enables parallel restoration operations, and significantly improves restoration speed while reducing storage costs through efficient use of existing host storage resources.
3Reliability
If mobile hosts are excluded from backup participation due to frequent disconnections, then backup reliability is improved, but available storage capacity and system versatility deteriorate
Solution Approach 1:
The patent implements a dynamic peer-to-peer backup system where mobile hosts can dynamically join and leave the backup network without compromising overall system reliability. The distributed architecture allows the system to adapt to changing host availability - when mobile hosts are connected, they contribute storage capacity; when disconnected, other hosts continue to maintain backup data. This dynamic participation increases total available storage capacity and system versatility while maintaining reliability through redundancy.
4Quantity of substance
If Single Instance Storage techniques are used to reduce storage pool size, then storage efficiency is improved, but restoration complexity and processing overhead increase
Solution Approach 1:
The patent uses fingerprint-based identification to recognize duplicate data segments across the peer network and stores only unique segments, implementing Single Instance Storage principles. Each data segment is assigned a fingerprint that serves as its unique identifier. When restoring data, the system retrieves segments based on fingerprints from the distributed hash table, simplifying the restoration process by directly locating needed segments without complex deduplication processing during restoration.
Data Source
AI summary
A system and method for storing data in a peer-to-peer network. A computer system includes interconnected hosts configured to store data segments. A first host stores a first subset of the data segments received from other hosts. The first host maintains a portion of a distributed hash table corresponding to the first subset of data segments and de-duplicates the first subset of the data segments against the remaining data segments. The distributed hash table comprises entries corresponding to the data segments, each entry including a data segment fingerprint that unambiguously identifies the corresponding data segment. The first host selects and joins a group of hosts that maintains the distributed hash table. The first host conveys data to the selected group indicating its availability to own additional entries in the distributed hash table.