Data Cluster Duplication via Read Write Temperature Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data cluster duplication techniques face challenges in identifying suitable candidates, managing overhead and synchronization, especially for frequently updated data clusters, and maintaining storage capacity, leading to performance bottlenecks and hotspots in computer systems.
Innovation Solution
A method is introduced to select data cluster duplication candidates based on read and write temperature thresholds, where clusters with high read temperatures and low write temperatures are identified for duplication, allowing for efficient routing of requests to the least busy storage devices and dynamic management of duplicates to optimize storage capacity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple physical copies of data clusters are created to ease congestion, then data access performance is improved, but storage capacity is reduced and synchronization overhead increases
Solution Approach 1:
The patent changes the parameter of duplication selection from arbitrary or manual to temperature-based automatic selection. By monitoring read temperature (access frequency) and write temperature (update frequency) of data clusters, the system dynamically identifies suitable candidates for duplication - specifically those with high read temperature and low write temperature - thereby optimizing storage capacity utilization while maintaining performance benefits
Solution Approach 2:
The patent creates physical copies of selected data clusters to establish multiple data paths. By routing read requests to the least busy storage device containing a copy, the system improves data access performance and eliminates bottlenecks without duplicating all data, thus preserving storage capacity
2Productivity
If multiple physical copies of data clusters are created, then data access performance is improved, but synchronization overhead and complexity increase
Solution Approach 1:
The patent introduces temperature parameters (read temperature and write temperature) to automatically identify suitable duplication candidates. Clusters with high read temperature and low write temperature are selected, which naturally reduces synchronization overhead since these clusters are read frequently but written rarely, minimizing the need to update multiple copies
Solution Approach 2:
The patent creates physical copies only for selected clusters rather than all clusters. By routing read requests to copies and write requests to the original, the system improves performance while avoiding the complexity of synchronizing writes across multiple copies
3Productivity
If data clusters are duplicated frequently, then access contention is reduced, but storage capacity availability decreases
Solution Approach 1:
The patent uses temperature parameters to dynamically determine duplication candidates. By selecting only clusters with high read temperature and low write temperature, the system reduces access contention for frequently read data while preserving storage capacity for clusters that don't benefit from duplication
Solution Approach 2:
The patent applies duplication selectively to specific data clusters based on their access characteristics rather than uniformly to all clusters. This localized approach ensures storage capacity is allocated efficiently - duplicated for read-intensive clusters, preserved for write-intensive or infrequently accessed clusters
Data Source
AI summary
A method of selecting candidates for data cluster duplication that can be used as an alternative or as an addition to existing duplication techniques. The method determines a read temperature of a data cluster. If the read temperature of the data cluster exceeds a threshold value, a write temperature of the data cluster is determined. If the write temperature of the data cluster is below a threshold value the cluster is selected for duplication.