Storage Device I/O Cost Normalization for Mixed Workload Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional performance metrics, such as IOPS and throughput, are inadequate for evaluating the performance of storage devices under mixed workloads with varying ratios of read and write requests and different request sizes, leading to inaccurate resource management and reduced efficiency in cloud and distributed storage systems.
Innovation Solution
A system that calculates a new performance metric, 'virtual I/O operations per second' (VIOPS), normalizing the cost of I/O requests independent of access patterns and types, allowing for more accurate allocation and scheduling of resources based on the storage device's capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional performance metrics (IOPS or throughput) are used to evaluate storage device performance, then the metrics are simple to calculate and understand, but they cannot accurately evaluate performance under mixed workloads with varying read/write ratios and request sizes
Solution Approach 1:
The patent transforms the conventional IOPS metric into a normalized metric by introducing a cost factor parameter that accounts for different I/O operation characteristics. Each I/O request is assigned a cost factor based on its type (read/write), size, and access pattern, converting diverse I/O operations into a unified normalized IOPS metric that accurately reflects storage device performance under mixed workloads
Solution Approach 2:
The patent introduces a normalization layer between the raw I/O requests and the performance metric calculation. This intermediary mechanism uses cost factors to translate various I/O operations (different types, sizes, and patterns) into a common normalized unit, enabling accurate performance comparison without directly complicating the metric calculation
2Adaptability or versatility
If conventional IOPS metrics are used without normalization, then the calculation is straightforward, but the metric cannot account for asymmetric read and write performance characteristics of NAND media and garbage collection mechanisms
Solution Approach 1:
The patent applies different cost factor values to different I/O operation characteristics (read vs. write, different sizes, different access patterns). This local differentiation allows the metric to capture the asymmetric performance characteristics of NAND media and garbage collection mechanisms while maintaining a unified calculation framework
Solution Approach 2:
The cost factors are dynamically determined based on the specific characteristics of each I/O request and the current state of the storage device. This dynamic adjustment enables the metric to adapt to varying workload conditions and accurately reflect performance under different scenarios
3Measurement precision
If normalized cost factors are applied to all I/O requests, then the performance metric becomes independent of access patterns and request types, but the determination of appropriate cost factor values requires extensive testing across multiple workload scenarios
Solution Approach 1:
The patent performs cost factor calibration in advance through systematic testing across multiple workload scenarios before deploying the performance metric in production. This preliminary action establishes the normalized cost factors needed for accurate performance measurement, avoiding the need for extensive testing during operational phases
Data Source
AI summary
One embodiment facilitates measurement of a performance of a storage device. During operation, the system determines a normalized cost for an I/O request, wherein the normalized cost is independent of an access pattern and a type of the I/O request, wherein the normalized cost is indicated by a first number of virtual I/O operations consumed by the I/O request, and wherein a virtual I/O operation is used as a logical unit of cost associated with physical I/O operations. The system identifies a performance metric for the storage device by calculating a second number of virtual I/O operations per second which can be executed by the storage device. The system allocates incoming I/O requests to the storage device based on the performance metric, e.g., to satisfy a Quality of Service requirement, thereby causing an enhanced measurement of the performance of the storage device.