Dual-NVM Memory Management for Predictable Mixed-Workload Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage solutions, such as SSDs, face challenges in providing consistent data access latency and performance in mixed read-write scenarios, especially when used as memory tiering devices over high-speed interfaces like Compute Express Link (CXL), due to unpredictable performance variations.
Innovation Solution
A storage device utilizing a mix of volatile memory (e.g., DRAM) and non-volatile memories (NVMs) operates NVMs in separate read and write modes to handle read and write requests separately, ensuring predictable performance by avoiding read/write conflicts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single NVM device is used to handle both read and write requests, then device complexity is reduced, but data access latency becomes unpredictable due to read/write conflicts
Solution Approach 1:
The patent divides the single NVM device into multiple separate NVM devices, where each device is dedicated to handling either read requests or write requests. This segmentation eliminates read/write conflicts that occur in single-device architectures, ensuring predictable data access latency while maintaining manageable system complexity through clear functional separation.
2Reliability
If NVMs operate in separate read and write modes, then data access latency becomes predictable, but device complexity increases due to multiple NVM devices
Solution Approach 1:
The patent assigns different operational characteristics to different NVM devices based on their specific functions. Read-optimized NVM devices are configured for high-speed read operations while write-optimized devices are configured for efficient write operations. This local quality differentiation allows each device to excel at its designated task, providing predictable overall system performance while managing complexity through specialized design.
3Productivity
If multiple NVM devices are used for read and write operations, then performance under mixed workloads improves, but cost increases
Solution Approach 1:
The patent implements a dynamic workload distribution mechanism that intelligently directs read and write requests to appropriate NVM devices based on current system state and workload characteristics. This dynamic approach maximizes the utilization of each NVM device, ensuring high performance under mixed workloads while optimizing the total number of devices required by adapting to changing operational demands.
Data Source
AI summary
Systems and methods for managing memory devices are disclosed. A device may include a controller, a first non-volatile memory (NVM) device, and a second NVM device. The controller may be configured to: detect a first condition; set, based on the first condition, the first NVM device for operating in a first operation mode and the second NVM device for operating in a second operation mode; receive a request from a computing device; and select one of the first NVM device or the second NVM device based on the request being respectively one of a first type or a second type, wherein the first NVM device is configured take a first action according to the first operation mode, and the second NVM device is configured take a second action according to the second operation mode.