Storage Modeler for Dynamic SLO Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current planner engines in storage systems struggle to dynamically consider new information when predicting storage metrics, leading to inaccuracies due to static evaluation functions and complex cluster environments, making it difficult to ensure Service-Level Objectives (SLOs) are met without significant reconfiguration.
Innovation Solution
A modeler-training engine is introduced to train a mapping function using machine learning techniques, summarizing input metrics from multiple workloads to predict output metrics, allowing for dynamic adaptation and improved accuracy in predicting storage system performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If static evaluation functions are used in planner engines, then device complexity is reduced, but measurement precision of storage metrics deteriorates
Solution Approach 1:
The patent applies dynamics by transitioning from static evaluation functions to dynamic machine learning models that continuously learn and adapt to changing storage cluster environments. The modeler retrains evaluation functions based on new information and changing conditions, enabling the system to dynamically adjust its prediction accuracy without requiring complete reconfiguration of the planner engine architecture.
Solution Approach 2:
The patent implements parameter changes by using machine learning techniques to transform fixed evaluation parameters into adaptive parameters that evolve based on training data. The modeler modifies evaluation function parameters through continuous learning from storage cluster metrics, allowing the system to maintain high measurement precision while keeping the overall device complexity manageable through automated parameter adjustment.
2Adaptability or versatility
If machine learning techniques are implemented for dynamic adaptation, then adaptability of the system improves, but device complexity increases
Solution Approach 1:
The patent applies self-service by implementing automated model training and evaluation function generation without requiring manual intervention. The modeler automatically retrains evaluation functions using stored training data and new information from the storage cluster, enabling the system to adapt dynamically while keeping operational complexity low through self-service automation. The planner engine simply invokes the generated evaluation functions without needing to understand the underlying machine learning complexity.
3Measurement precision
If comprehensive training data from multiple workloads is processed, then measurement precision of predictions improves, but loss of time for data processing increases
Solution Approach 1:
The patent implements preliminary action by pre-processing and storing training data from multiple workloads in advance. The system collects and stores input metrics and output metrics from various workloads before they are needed for evaluation. This allows the modeler to quickly retrieve and process pre-organized training data when retraining evaluation functions, reducing the time loss during actual prediction tasks while maintaining high measurement precision through comprehensive data coverage.
Data Source
AI summary
Described herein is a system and method for dynamically managing service-level objectives (SLOs) for workloads of a cluster storage system. Proposed states/solutions of the cluster may be produced and evaluated to select one that achieves the SLOs for each workload. A planner engine may produce a state tree comprising nodes, each node representing a proposed state/solution. New nodes may be added to the state tree based on new solution types that are permitted, or nodes may be removed based on a received time constraint for executing a proposed solution or a client certification of a solution. The planner engine may call an evaluation engine to evaluate proposed states, the evaluation engine using an evaluation function that considers SLO, cost, and optimization goal characteristics to produce a single evaluation value for each proposed state. The planner engine may call a modeler engine that is trained using machine learning techniques.