Multi-layer Consistent Hashing for Distributed Storage Node Failure
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Distributed storage systems face challenges in fault tolerance and load balance, particularly in large-scale applications where server failures can lead to data inconsistencies and uneven data distribution, affecting system availability and reliability.
Innovation Solution
A distributed data storage system utilizing multiple-layers consistent hashing, which includes storage nodes, management nodes for mapping and redundancy, and monitor nodes for state maintenance, ensures fault tolerance and load balance through heartbeat messages and algorithms for primary and secondary node selection, failure detection, and data rebalancing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple servers are used to store data together, then storage capacity and reliability are improved, but the probability of server failures increases and data inconsistencies may occur
Solution Approach 1:
The patent segments data into multiple pieces and stores them across different servers in a distributed manner. Each piece of data is divided into multiple segments that can be independently stored and retrieved, reducing the impact of any single server failure while maintaining overall system reliability.
Solution Approach 2:
The patent employs consistent hashing with virtual nodes to dynamically adjust data distribution parameters. When server failures occur, the system recalculates hash values and redistributes data segments to available nodes, changing the allocation parameters adaptively to maintain reliability despite node failures.
2Reliability
If data is divided into multiple pieces stored on different servers, then fault tolerance is improved, but data inconsistencies between replicas may occur
Solution Approach 1:
The patent implements a feedback mechanism where the system continuously monitors the state of data replicas across nodes. When inconsistencies are detected or nodes fail, the system triggers rebalancing operations that coordinate data synchronization, ensuring all replicas remain consistent through iterative feedback loops.
Solution Approach 2:
The patent performs preliminary data replication and consistency verification before failures occur. Data is pre-segmented and distributed to multiple nodes with consistent hashing rules established in advance, so that when failures occur, the system can quickly recover without data inconsistency by following pre-defined rebalancing procedures.
3Quantity of substance
If the number of servers increases, then storage capacity is improved, but load balance and data routing efficiency deteriorate
Solution Approach 1:
The patent creates a universal data routing layer using consistent hashing that works independently of the number of nodes. The hashing function provides a universal mapping mechanism that scales with the system, allowing data to be routed to any node in the distributed system without requiring complex routing logic, thus maintaining efficiency as capacity increases.
Solution Approach 2:
The patent introduces virtual nodes as an intermediary layer between physical servers and data segments. These virtual nodes act as mediators that absorb the complexity of data routing, allowing the system to scale storage capacity by adding physical nodes while maintaining efficient routing through the stable virtual node layer that uses consistent hashing for deterministic data placement.
Data Source
AI summary
Disclosed is a systems and methods of distributed data storage using multi-layers consistent hashing comprising: a plurality of storage nodes providing data storage and redundancy protection; a plurality of management nodes maintaining the properties of the storage nodes and mapping information from virtual groups to storage nodes; a plurality of monitor nodes maintaining the state of storage nodes and handling the changes of states of storage nodes including joining, decommissioning and failure; and one or more clients providing entries for applications or users to access the storage system. The storage nodes is in a hierarchical tree arrangement, and each storage node in each layer of the tree is allocated with a plurality of identities and configured for remaining hash space with consistency. Instead of sharing one hash space among all storage nodes, there are a plurality of hash spaces kept consistent in each layer of the storage hierarchical tree.