Storage Node Grouping for Load Balancing and Data Protection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing data storage systems face instability and overload when a node fails, leading to poor predictability of data loss and resource utilization, causing instability and increased complaints from users.

Innovation Solution

A data storage method where all devices constitute a single data pool, with a management server that polls and selects a group of devices for data storage, redistributing data to prevent overload and ensuring decentralized storage across multiple nodes, thereby sharing access traffic and maintaining stability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If all access traffic is taken over by remaining nodes in a cluster when a node fails, then data protection is achieved, but node stability deteriorates due to overload

Engineering Contradiction:
Improvedata protectionVSAvoidnode stability
Core Design Contradiction:
ReliabilityVSStability of the object's composition

Solution Approach 1:

The patent segments the cluster into multiple sub-clusters, where each sub-cluster independently manages its own data and has dedicated backup nodes. When a node fails, only the affected sub-cluster experiences load redistribution, while other sub-clusters remain unaffected. This segmentation prevents system-wide overload and maintains node stability while still achieving data protection within each sub-cluster.

Inventive Principle:
Principle #1Segmentation

2Reliability

If data is stored in clusters with redundant nodes, then data protection is improved, but resource utilization deteriorates due to low efficiency

Engineering Contradiction:
Improvedata protectionVSAvoidresource utilization
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent implements local quality by allowing different sub-clusters to have different data storage configurations and backup strategies tailored to their specific requirements. Each sub-cluster can optimize its resource allocation independently, improving overall resource utilization while maintaining data protection. The system assigns backup nodes locally within sub-clusters rather than requiring global redundancy, reducing wasted resources.

Inventive Principle:
Principle #3Local quality

3Reliability

If multiple nodes store identical data in a cluster, then data protection is achieved, but predictability of data loss deteriorates

Engineering Contradiction:
Improvedata protectionVSAvoidpredictability of data loss
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent implements feedback mechanisms that continuously monitor the status of nodes and data storage across sub-clusters. This feedback enables the system to accurately track which data resides on which nodes and to predict the impact of potential failures. The management server uses this information to provide deterministic calculations of data loss scenarios, improving predictability while maintaining data protection through redundant storage in sub-clusters.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP2202921B1A data storage method, a management server, a storage equipment and system
Publication Date: 2013.03.27 CHINA MOBILE COMM GRP CO LTD
  • EP2202921B1 patent drawingFigure 1~2
  • EP2202921B1 patent drawingFigure 3~4
  • EP2202921B1 patent drawingFigure 5

AI summary

A data storage method, a management server, a storage equipment and system are provided in the present invention, in which the data storage method includes that all n equipments used for storing data constitute a data pool; when needing to store data, all equipments of the data pool are inquired one by one to select m equipments as one group, and then the said data are stored in the m equipments of the selected group respectively, herein m is larger than 1 and smaller than n. The embodiments of the present invention can resolve the problems that the load of the other nodes is increased and the other nodes are insecure because some node is failed, the availability of every node is low and the predictability of every node is worse in the current data storage manner; the embodiments of the present invention can average the load of the equipment and insure the higher stability of the node when some node is failed, and improve the source availability and predictability of the node.