Adaptive Block Allocation Mechanism for Storage Slices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional block allocation mechanisms in storage systems do not effectively account for various factors affecting storage performance, leading to inefficiencies, especially in tiered storage systems where faster devices may go underutilized while slower devices with greater capacity are overutilized.
Innovation Solution
An adaptive block allocation mechanism that assigns scores to storage slices based on multiple characteristics, such as speed, capacity, and type of storage media, to optimize allocation decisions and improve storage performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional block allocation mechanisms (round-robin or most-space available) are used, then allocation simplicity is maintained, but storage performance deteriorates due to improper utilization of tiered storage devices
Solution Approach 1:
The system pre-calculates and stores scores for each storage slice based on characteristics like IOPS, latency, and capacity before allocation requests arrive. When a allocation request comes in, the system simply queries the pre-computed scores and selects the best slice, avoiding complex real-time calculations and achieving both high performance and operational simplicity
Solution Approach 2:
The patent replaces traditional mechanical allocation approaches (round-robin scheduling, most-space-available heuristics) with an information-based scoring system. Instead of relying on simplistic rules, the system uses computed scores that reflect actual storage characteristics and performance metrics, substituting mechanical allocation logic with intelligent decision-making based on pre-analyzed data
2Quantity of substance
If most-space available approach is used in tiered storage, then large capacity devices are充分利用, but faster storage devices go underutilized
Solution Approach 1:
The system assigns different characteristic weights to different storage slices based on their specific properties. Fast storage devices (SSD, flash) receive higher weights for IOPS and latency characteristics, while large capacity devices (HDD) receive higher weights for capacity metrics. This localized quality adjustment ensures each device type is evaluated according to its strengths, optimizing both speed and capacity utilization
Solution Approach 2:
The system dynamically adjusts the scoring parameters and characteristic weights based on the specific storage slice being evaluated. By changing the relative importance of different parameters (IOPS, latency, capacity, end-to-end latency) for different storage types, the system ensures that fast storage is selected when speed is critical while still utilizing large capacity storage when appropriate
3Productivity
If thin logical units allocate storage on demand, then storage efficiency is improved, but allocation performance optimization becomes more challenging
Solution Approach 1:
The system pre-computes scores for all available storage slices based on their characteristics and current state before allocation requests arrive. This preliminary action creates a ready-to-use ranking system that simplifies on-demand allocation decisions, maintaining storage efficiency while reducing the complexity of real-time allocation decisions
Solution Approach 2:
The patent introduces a scoring system as an intermediary between the thin logical unit allocation model and the physical storage devices. Instead of directly evaluating multiple device characteristics for each allocation request, the scoring system acts as a mediator that has already processed and ranked the options, simplifying the allocation decision while maintaining optimization
Data Source
AI summary
Systems, methods, and computer readable media for an adaptive block allocation mechanism are disclosed. According to one aspect, a method for allocating slices of storage in a storage medium based on scores is disclosed. The method occurs at a storage processor for controlling access to a storage medium divided into a plurality of logical slices of storage. The method includes determining one or more scores associated with one or more slices. Determining one or more scores includes combining plural dissimilar characteristics of the slice into a single value. The method further includes receiving a slice allocation request, and in response to the slice allocation request, selecting a slice for allocation using the scores.