Distributed Storage Data Replication via Access Pattern Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Distributed storage systems in mobile communications networks face challenges with data replication, as they require large storage resources and high communication demands, especially when dealing with large data sets and network latency issues due to centralized storage systems.
Innovation Solution
A method for data replication in a distributed storage system that anticipates and pre-emptively stores data at a storage location based on identified access patterns, by determining further data to be stored at a first location and replicating it from a second location, thereby reducing storage resources and latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If data replication employs a master/slave model to ensure data availability at geographically separated locations, then data accessibility is improved, but storage resource requirements and communication demands increase significantly
Solution Approach 1:
The system performs preliminary actions by identifying access patterns and pre-replicating data to edge storage locations before actual access requests occur. This anticipatory replication reduces the need for full data sets at all locations, optimizing storage resource usage while maintaining data accessibility.
Solution Approach 2:
The patent applies local quality by customizing the data replication strategy for each edge storage location based on its specific access patterns. Different locations receive different subsets of data tailored to their local access characteristics, reducing overall storage requirements while ensuring each location has the data it actually needs.
2Reliability
If all instances store identical replicas of large data sets to ensure data consistency, then data reliability is improved, but communication link demands and storage costs increase
Solution Approach 1:
The system applies partial action by replicating only the necessary portions of data to each edge location based on access patterns, rather than complete data sets. This reduces communication overhead and storage requirements while maintaining sufficient data consistency for local operations.
Solution Approach 2:
The master node performs preliminary analysis of access patterns and proactively replicates only the data subsets that will be needed at each edge location, reducing unnecessary communication and storage while maintaining data reliability for actual access scenarios.
3Device complexity
If data is stored centrally in a single UDR to simplify storage management, then system complexity is reduced, but network latency increases for edge access
Solution Approach 1:
The patent segments the centralized storage system into a master node that manages replication decisions and multiple edge storage locations that hold local data subsets. This segmentation reduces network latency for edge access while the master node maintains simplified management through access pattern-based replication strategies.
4Speed
If data replication pre-loads data at edge locations to reduce network latency, then access speed is improved, but storage resource requirements at each instance increase
Solution Approach 1:
The system applies local quality by tailoring the data replication to each edge location's specific access patterns. Each location stores only the data subsets that are relevant to its local access characteristics, reducing storage provisions while maintaining fast access speed for locally-relevant data.
Solution Approach 2:
The system performs preliminary analysis of access patterns to determine which data subsets should be pre-loaded at each edge location. This anticipatory approach ensures that only necessary data is stored locally, achieving fast access speed without excessive storage provisions.
Data Source
AI summary
A method of data replication in a distributed storage system of a mobile communications network is disclosed. The distributed storage system comprises a plurality of geographically separated storage locations. The method comprises, at a first storage location: receiving a first request for data stored in the distributed storage system; determining, based on the first request, further, different, data to be stored at the first storage location; determining a second, different, storage location, of the plurality of storage locations, at which the determined further data is stored; sending, to the determined second storage location, a replication request for replicating the determined further data; receiving, in response to the replication request, the further data; and storing the further data such that the further data is replicated at the first storage location in advance of a further request for the further data being received at the first storage location. Apparatus is also disclosed.