Selective Data Migration for Storage Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data migration systems require moving large portions of data when only small portions are needed, leading to inefficient use of storage resources, especially when data needs to be optimized across different geographic locations for various criteria such as cost and proximity to computing resources.
Innovation Solution
The system subdivides logical devices into subsections, where each subsection corresponds to a specific address range mapped to physical devices, allowing data to be copied from one physical device to another while maintaining some data indefinitely on the original device, using automated algorithms to determine which data to move based on expected activity levels, and presenting data to hosts independently of its physical location.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If all data for a logical unit is moved between physical storage locations, then data can be optimized for different criteria (cost, proximity to computing resources) at different times, but unnecessary storage resources are consumed by moving entire datasets when only small portions are needed
Solution Approach 1:
The patent divides a logical unit into multiple data portions (e.g., hot data, warm data, cold data) and independently manages the physical location of each portion. This allows selective migration of only specific data portions between storage locations based on access patterns and optimization criteria, rather than moving entire datasets. The segmentation enables granular control over data placement while maintaining logical unit integrity.
2Ease of operation
If data is moved manually or by automated process before payroll processing, then data can be available at the correct location for processing, but the entire employee data set must be moved even if only a small portion is needed
Solution Approach 1:
The patent extracts only the specific data portions required for payroll processing from the larger employee data set and moves them to the appropriate location. By identifying and extracting only the necessary data (e.g., active employee records needed for current payroll cycles) rather than moving all employee data, the system improves migration efficiency while ensuring data availability for processing.
3Ease of operation
If data is presented to hosts independently of physical location, then hosts can access data transparently, but the underlying physical storage optimization becomes complex
Solution Approach 1:
The patent introduces a data management system as an intermediary layer between hosts and physical storage devices. This mediator handles the complexity of tracking and managing the physical locations of data portions while presenting a simplified, location-transparent interface to hosts. The intermediary maintains mappings between logical units and their distributed physical portions, absorbing the management complexity away from hosts and enabling transparent access.
Data Source
AI summary
The system described herein allows for picking data to be moved from one physical location to another using certain criteria, such as expected activity level, to achieve load balancing or load optimization. Migrating data for a logical device includes subdividing the logical device into subsections, where each of the subsections corresponds to a particular address range of the logical device that is mapped to a physical device containing corresponding data and copying data for a first subsection from a first physical device to a second, different, physical device, where data for a second subsection is maintained indefinitely on the first physical device after copying the data for the first subsection to the second physical device. The first physical device may be provided in a different geographic location than the second physical device.