Multi-Level Data Consolidation for Storage Utilization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage systems face inefficiencies in data consolidation, particularly when PLBs are highly utilized, as many old PLBs cannot be combined into new ones due to size constraints, leading to a scenario where nearly half of the storage space remains unused without consolidation occurring.
Innovation Solution
The technique consolidates data at multiple levels of granularity by selecting sets of whole PLBs and portions of PLBs based on utilization, tracking PLBs in queues arranged by utilization, and using PLB portions to fill residual space in new PLBs, allowing for efficient data consolidation even in highly utilized systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data consolidation combines old PLBs into new PLBs, then free PLBs are created and system health is maintained, but consolidation cannot occur when PLBs are highly utilized because combinations exceed new PLB capacity
Solution Approach 1:
The patent segments PLBs into two levels: whole PLBs and portions of PLBs. Whole PLBs are combined based on utilization thresholds, and remaining space is filled with portions from donor PLBs. This segmentation allows consolidation to proceed even when whole PLBs alone cannot fit, thereby resolving the contradiction between maintaining consolidation efficiency and adapting to high utilization scenarios.
Solution Approach 2:
The patent applies different consolidation strategies to different parts of the storage system based on local utilization characteristics. PLBs with utilization below a threshold are combined as whole units, while PLBs with higher utilization contribute portions to fill residual space. This local quality approach enables the system to adapt consolidation behavior to local conditions, resolving the contradiction between consolidation efficiency and adaptability to varying utilization levels.
2Ease of manufacture
If whole PLBs are combined into new PLBs, then consolidation is simple to implement, but nearly half of storage space remains unused when all PLBs are just over 50% utilized
Solution Approach 1:
The patent divides the consolidation process into two stages: first combining whole PLBs that meet utilization criteria, then filling remaining space with portions from donor PLBs. This segmentation maintains the simplicity of whole-PLB combining while adding a second stage to utilize otherwise wasted space, thereby resolving the contradiction between process simplicity and space utilization.
Solution Approach 2:
The patent performs partial consolidation by taking portions from donor PLBs rather than requiring complete PLB combinations. This partial action allows the system to consolidate space that would otherwise remain unused, resolving the contradiction between maintaining simple whole-PLB processes and maximizing usable storage space.
Data Source
AI summary
A technique consolidates data at multiple levels of granularity, the levels including a first level based on whole PLBs (physical large blocks) and a second level based on portions of donor PLBs. The technique further includes tracking PLBs in multiple PLB queues arranged based on storage utilization of the PLBs, and tracking PLB portions in multiple portion queues arranged based on storage utilization of the portions. When consolidating data to create a new PLB, a set of whole PLBs is selected, based on utilization, from the PLB queues, and a set of portions of donor PLBs is selected, based on utilization, from the portion queues. The selections are performed such that the total data size of the selected whole PLB(s) and the selected portion(s) fit within the new PLB.