Storage Resource Allocation via Fuzzy Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional storage systems face challenges in efficiently and timely deploying resources to meet workload demands, leading to resource wastage and the need for constant administrator intervention due to the lack of self-adjustment capabilities and centralized operation.
Innovation Solution
A method that sets state-action and action-reward fuzzy rules to dynamically adjust storage resources based on observed and predicted deviations from Service Level Agreements, utilizing an experience matrix to learn and optimize resource allocation over time, allowing the system to automatically deploy resources without human interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If administrators assign permanently many resources to meet SLA requirements during sudden peaks, then workload performance requirements are met, but resource wastage occurs as resources stand by during low-demand periods
Solution Approach 1:
The patent implements dynamic resource allocation by continuously monitoring workload characteristics and performance metrics, then adjusting resource allocation in real-time based on current demands and predicted future workloads. This replaces static permanent allocation with adaptive dynamic allocation that matches resources to actual needs.
Solution Approach 2:
The system performs preliminary actions by predicting future workload patterns and proactively allocating resources before performance degradation occurs. The performance model anticipates upcoming resource needs and triggers pre-emptive resource allocation to maintain SLA compliance.
2Ease of operation
If a centralized controlling server is used to manage resource deployment, then resource configuration is centralized and controllable, but the controlling server becomes a bottleneck when workloads increase
Solution Approach 1:
The patent segments the centralized controlling server into multiple distributed resource allocation nodes that operate autonomously. Each node manages specific storage resources and workloads independently, eliminating the single-point bottleneck while maintaining coordinated resource allocation through shared performance models.
3Adaptability or versatility
If administrators manually adjust rules and policies in the rule-base engine, then the system can adapt to different workloads, but constant administrator intervention is required and adjustments may be delayed
Solution Approach 1:
The patent implements self-service through an automated performance model that continuously learns from actual workload behavior and automatically generates resource allocation decisions. The system monitors its own performance, identifies optimization opportunities, and executes adjustments without administrator intervention, enabling continuous adaptation.
Solution Approach 2:
The system establishes continuous feedback loops where performance metrics from actual workload execution are fed back into the performance model. The model uses this feedback to refine its predictions and automatically adjust resource allocation, creating a closed-loop adaptive system that responds to changing conditions in real-time.
Data Source
AI summary
A method for deploying storage system resources with learning of workloads applied to a storage system is disclosed. The method utilizes state-action fuzzy rules and action-reward fuzzy rules for the storage system to learn a best dynamic deployment of resources from the parameters of workloads. No more human interference is required. It reduces work loading of the storage system administrator and enhances performance of the storage system.