Data Storage Response Time Prediction via Workload Simulation

Solution Overview

Problem

Designing and configuring data storage systems is complex and time-consuming, and existing methods do not effectively predict performance until after the system is implemented, leading to potential unsatisfactory performance and additional costs to improve it.

Innovation Solution

A method that defines a data storage system by processing circuitry, applying a workload with specified IOPS requirements and IO request percentages to determine response time, considering IO request queue length, service time, read cache hit response time, and write cache hit response time.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If storage system design and configuration is performed manually after implementation, then system can be built, but performance cannot be predicted accurately and additional time and expense are required to improve performance

Engineering Contradiction:

Improveperformance prediction accuracyVSAvoiddesign and configuration time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary action by performing workload simulation and performance prediction before the storage system is actually designed and configured. The system defines a workload model with IOPS requirements and IO request percentages, then uses this to calculate expected response times and queue lengths in advance, allowing designers to optimize configuration parameters before implementation.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent creates a virtual copy or model of the storage system's operational behavior through workload simulation. By replicating the system's characteristics in a simulated environment, designers can predict performance metrics without building the actual physical system, thus avoiding time-consuming manual testing and configuration adjustments.

Inventive Principle:

Principle #26Copying

2Reliability

If storage system is designed without accurate performance prediction capability, then design process is simpler, but performance may be unsatisfactory requiring additional improvement efforts

Engineering Contradiction:

Improveperformance satisfactionVSAvoiddesign and configuration complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements feedback mechanisms by calculating expected response times and queue lengths based on workload characteristics and system configuration parameters. These predicted performance metrics provide feedback to designers, allowing them to adjust configuration settings to meet performance requirements, thereby ensuring satisfactory system performance without unnecessary complexity.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent utilizes parameter changes by allowing designers to modify configuration parameters such as queue length thresholds, service time estimates, and workload mix ratios. By adjusting these parameters within the simulation model, designers can optimize system performance for different scenarios without implementing complex physical changes, thus achieving reliable performance with manageable design complexity.

Inventive Principle:

Principle #35Parameter changes

3Productivity

If manual performance optimization is performed after system implementation, then system can be built initially, but additional time and expense are required to improve performance

Engineering Contradiction:

ImprovethroughputVSAvoidperformance improvement time

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by performing workload simulation and performance prediction before the storage system is actually designed and configured. The system defines a workload model with IOPS requirements and IO request percentages, then uses this to calculate expected response times and queue lengths in advance, allowing designers to optimize configuration parameters before implementation.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent creates a virtual copy or model of the storage system's operational behavior through workload simulation. By replicating the system's characteristics in a simulated environment, designers can predict performance metrics without building the actual physical system, thus avoiding time-consuming manual testing and configuration adjustments.

Inventive Principle:

Principle #26Copying

Data Source

PatentUS10592137B1Method, apparatus and computer program product for determining response times of data storage systems

Publication Date: 2020.03.17 EMC IP HLDG CO LLC

AI summary

Techniques are disclosed for use in determining response times of data storage systems. In one embodiment, there is disclosed a method. The method comprises defining a data storage system being designed. The method also comprises defining a first workload for the data storage system. The first workload including a first IOPS (input-output operations per second) requirement and respective percentages of read and write IO (input-output) requests. The method also comprises applying the first workload to the data storage system, thus defining a IO request queue length. The method further comprises determining a response time for handling an IO request at the data storage system, wherein the said determination is based on the IO request queue length, the respective percentages of read and write IO requests, a service time relating to servicing of an IO request by a data storage device of the data storage system, a read cache hit response time and a write cache hit response time associated with a cache of the data storage system.