Storage Bandwidth Control via Predictive Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional bandwidth adjustment methods in storage systems are slow to respond to rapid changes in input-output operations due to stepwise adjustments, making it difficult to achieve target response times and prioritize tasks effectively.
Innovation Solution
A storage management device that obtains performance information, calculates target bandwidth based on standby times, and adjusts bandwidth dynamically to achieve target response times by expanding or narrowing bandwidth limits, allowing for more expeditious adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If bandwidth is adjusted in a stepwise manner by comparing actually-measured response time and target response time, then the storage system can achieve the target response time, but it takes time to reach the appropriate bandwidth and cannot follow unexpected or rapid changes in input-output operations
Solution Approach 1:
The patent applies preliminary action by pre-calculating the relationship between bandwidth and response time through modeling. Instead of adjusting bandwidth stepwise based on actual measurements, the system pre-establishes a model that directly maps target response times to required bandwidth values. This allows the system to jump directly to the appropriate bandwidth setting without iterative adjustments, resolving the contradiction between measurement precision and adjustment speed.
Solution Approach 2:
The patent replaces the mechanical stepwise adjustment system with a computational modeling approach. By substituting the physical trial-and-error adjustment mechanism with a mathematical model that directly calculates optimal bandwidth from target response time, the system achieves both precise response time control and rapid bandwidth adjustment without the delays inherent in stepwise mechanical adjustment.
2Reliability
If conventional bandwidth adjustment methods are used, then the system can control bandwidth, but it is difficult to respond to unexpected or rapid changes in input-output operations
Solution Approach 1:
The patent applies dynamics by creating a flexible bandwidth adjustment system based on predictive modeling. Instead of rigid stepwise adjustment, the system dynamically calculates optimal bandwidth settings in advance using a model that predicts the relationship between bandwidth and response time. This allows the system to adapt rapidly to changing I/O conditions while maintaining reliable bandwidth control, as the model can quickly recalculate optimal settings without being constrained by stepwise adjustment limits.
3Manufacturing precision
If stepwise bandwidth adjustment is performed, then the system can approach target bandwidth, but adjusting the bandwidth takes time
Solution Approach 1:
The patent eliminates bandwidth adjustment time by performing preliminary action through pre-modeling the bandwidth-response time relationship. Instead of spending time on stepwise adjustments to reach target bandwidth, the system pre-establishes a mathematical model that directly determines the optimal bandwidth setting. This allows the system to achieve precise bandwidth control instantaneously by simply applying the model calculation results, completely eliminating the time loss associated with iterative stepwise adjustment.
Data Source
AI summary
A monitoring unit obtains the performance information of each of a plurality of volumes included in a storage device. A setting unit obtains a performance target value for each of one or more targeted volumes from among the plurality of volumes. Based on the performance information obtained by the monitoring unit and based on a first-type standby time that is attributed to bandwidth limitation of IO requests with respect to the targeted volumes which are obtained by the setting unit and which achieve the performance target value, a bandwidth managing unit obtains the bandwidth for the targeted volumes that achieve the performance target value and performs bandwidth limitation to change the bandwidth of the targeted volumes to the obtained bandwidth.