Storage Controller Hot-Cold Data Placement for Read Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage systems face challenges in optimizing data placement based on read frequency, leading to inefficiencies in average latency and throughput.
Innovation Solution
The proposed information processing system implements Hot Cold separation by categorizing data into Read Hot and Read Cold based on read frequency, and strategically places these data types in regions of the non-volatile memory with appropriate read speeds, using Gray coding and unbalanced coding methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If data is uniformly distributed across non-volatile memory regions, then storage capacity is maximized, but read latency increases due to slower access in lower-speed regions
Solution Approach 1:
The patent divides the non-volatile memory into multiple regions with different read speeds and selectively places data with different read frequencies in appropriate regions. Read Hot data is stored in high-speed regions while Read Cold data is stored in lower-speed regions, creating local optimization of read performance based on data characteristics.
Solution Approach 2:
The patent segments the non-volatile memory into multiple regions based on read speed characteristics and segments data into Read Hot and Read Cold categories based on read frequency. This segmentation enables differentiated storage strategies that optimize overall system performance.
2Loss of time
If Read Hot data is stored in high-speed regions, then read latency for frequently accessed data decreases, but the capacity of lower-speed regions is underutilized
Solution Approach 1:
The patent dynamically changes the storage location parameter for data based on read frequency parameters. By monitoring read operations and transitioning data between Read Hot and Read Cold categories, the system adapts data placement to changing access patterns, ensuring optimal utilization of both high-speed and lower-speed regions.
Solution Approach 2:
The patent implements dynamic data placement where data can transition between high-speed and lower-speed regions based on changing read frequency characteristics. This dynamic adaptation ensures that the storage system responds to varying workloads and maintains optimal performance.
3Loss of time
If data placement is optimized based on read frequency, then average read latency decreases, but system complexity increases due to additional management overhead
Solution Approach 1:
The patent implements self-service mechanisms where the storage system automatically monitors read frequencies and performs data migration between regions without requiring external intervention. The system self-adjusts data placement based on observed access patterns, reducing the need for complex external management while maintaining optimization.
Solution Approach 2:
The patent uses feedback from read operation monitoring to dynamically adjust data placement decisions. By continuously observing read frequencies and using this feedback to transition data between Read Hot and Read Cold categories, the system maintains optimal performance with automated decision-making.
Data Source
AI summary
A storage device includes a non-volatile memory, and a controller. The non-volatile memory includes a first region including first data readable with a number of voltage application times being equal to one or two, and a second region including second data readable with a number of voltage application times being equal to or more than three. The controller comprises read frequency information, and is configured to control writing, based on the read frequency information: a first segment with a first number of issuance times being equal to or more than a reference value to the first region or the second region; and a second segment with a second number of issuance times being less than the reference value to the second region.