Key-Value Store Replication via Physical Cache Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud service providers face challenges in replicating and maintaining database performance across data centers due to the proprietary nature of key-value stores, which are not easily scalable and do not handle failures or network uncertainty well, making it difficult to move user data between data centers while preserving database performance.
Innovation Solution
A method for key-value store mapping and replication that involves extracting relative data storage locations using data query caches to map data divisions and duplications within a target data center, and replicating these structures to a new data center, allowing for the construction of a replicated database tree structure with optimized record duplications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If key-value stores are replicated using conventional methods, then data can be moved between data centers, but database performance is lost due to abstracted metadata structures
Solution Approach 1:
The patent creates a copy of the physical storage distribution map from the source data center and applies it to the destination data center. Instead of copying complex metadata trees, the system extracts and replicates the essential storage location mappings that preserve performance characteristics.
Solution Approach 2:
The patent extracts only the critical storage location information from the complex key-value store structure. By taking out just the essential physical distribution data rather than the entire metadata structure, it simplifies replication while maintaining performance.
2Adaptability or versatility
If key-value stores are made scalable and cloud-ready, then service orientation is improved, but proprietary structures become hidden and difficult to replicate
Solution Approach 1:
The patent introduces an intermediary storage location map that bridges the gap between the hidden proprietary key-value store structure and the replication process. This intermediary structure makes the invisible physical distribution visible and replicable without exposing the underlying proprietary complexity.
3Quantity of substance
If data is distributed across multiple machines with metadata trees, then storage capacity is increased, but replication becomes complex requiring knowledge of division boundaries
Solution Approach 1:
The patent copies the storage location distribution pattern from the source to destination data center, preserving how data is divided and distributed across machines without needing to understand or replicate the complex boundary definitions.
Data Source
AI summary
Technologies are generally described for cache based key-value store mapping and replication. In some examples, key-value stores may be mapped for data structure replication through extraction of file breaks in an existing key-value store by iterating through the store and examining changes in cache addresses to detect jumps in address values. Specially formulated queries may be executed to return the values within an address range that spans a physical storage volume in order to recover full key-value sets that are physically grouped at a current data center including record duplicates. Such sets may be used to replicate or inform the key-value sets at a new location or in a new key-value store allowing construction of a replicated database tree structure complete with record duplications that develop as tables are optimized over time.