I/O Path Selection for Asymmetrical Storage Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional I/O scheduling methods fail to adapt to the asymmetry between read and write requests in storage systems that support de-duplication, leading to poor I/O service quality and inefficiency in bandwidth utilization.
Innovation Solution
A self-adaptive path selection policy that dynamically chooses between read and write paths based on historical I/O request performance, combined with a multi-queue framework to adjust queue depths for different data types, ensuring optimal service quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional I/O scheduling methods are used, then the system structure is simple, but the I/O service quality is poor and bandwidth utilization is inefficient
Solution Approach 1:
The patent segments the I/O scheduling system by creating separate read queues and write queues with independent path selection mechanisms. This segmentation allows each queue type to be optimized independently, improving I/O service quality while maintaining manageable system complexity through modular design
Solution Approach 2:
The patent implements dynamic path selection where the system can switch between read paths and write paths based on real-time I/O request characteristics and performance metrics. This dynamic adaptation enables the system to optimize bandwidth utilization and respond to changing workload patterns, significantly improving productivity
2Ease of operation
If read and write requests are processed using the same path, then the system is simple to manage, but lock contention increases and service quality deteriorates
Solution Approach 1:
The patent separates the processing paths into distinct read paths and write paths, eliminating lock contention that would occur if both request types shared the same path. This segmentation maintains ease of operation through clear path separation while significantly improving reliability by allowing concurrent read and write operations without interference
Solution Approach 2:
The patent applies different path selection strategies and performance optimizations specific to read operations and write operations. Each path type can be tuned locally for its specific characteristics, improving overall service quality while keeping the management approach straightforward through localized optimizations
3Productivity
If a single queue is used for all I/O requests, then the system structure is simple, but bandwidth utilization is inefficient in asymmetrical storage systems
Solution Approach 1:
The patent divides the single queue into separate read queues and write queues, allowing independent bandwidth allocation and optimization for each operation type. This segmentation enables efficient bandwidth utilization in asymmetrical storage systems where read and write patterns differ, while the modular queue structure keeps the complexity manageable
Solution Approach 2:
The patent dynamically adjusts queue parameters such as depth and priority based on workload characteristics and performance metrics. By changing these parameters adaptively, the system optimizes bandwidth utilization without requiring a completely complex queue structure, balancing productivity improvement with acceptable structural complexity
Data Source
AI summary
Embodiments of the present disclosure provide a method, an apparatus and a computer program product for processing an Input/Output (I/O) request. In one embodiment, a method for processing an I/O request comprises: receiving, at a host, a first I/O request of a first type for a storage device, wherein at least a first path configured for processing I/O requests of the first type and a second path configured for processing I/O requests of a second type exist between the host and the storage device; selecting, from the first path and the second path, a path for processing the first I/O request; and processing the first I/O request via the path.