Queuing Curve Analysis for Workload Performance Impact Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information processing systems face challenges in detecting performance impacting events across diverse workloads, leading to inefficiencies in resource provisioning due to varying demand profiles and shared resource usage among different workloads.
Innovation Solution
The system employs queuing curve analysis to model expected performance by analyzing historical data, generating models for specific workload types, and identifies performance impacting events using quality evaluation metrics, allowing for dynamic adjustment of compute, storage, and network resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If queuing curve analysis is used to model expected performance for performance impact detection, then detection accuracy is improved, but system complexity increases due to model generation and quality evaluation processes
Solution Approach 1:
The system performs preliminary actions by generating queuing curve models from historical performance data before actual performance impact detection occurs. These models are stored and reused for detecting performance impacts in similar workload scenarios, avoiding the need to generate models repeatedly and reducing real-time detection complexity.
Solution Approach 2:
The system creates simplified copies of performance characteristics through queuing curve models that represent expected performance behavior. Instead of analyzing complex raw performance data directly, the system uses these modeled representations to detect deviations indicating performance impacts, reducing computational complexity while maintaining detection accuracy.
2Measurement precision
If workload-specific models are generated for different workload types, then detection precision is improved, but processing time increases due to separate model generation and analysis
Solution Approach 1:
The system generates workload-specific queuing curve models in advance during periods when workloads are executing, storing these models for future detection use. This preliminary model generation avoids the need to create models during critical detection periods, reducing processing time delays while maintaining precision through customized workload models.
Solution Approach 2:
The system dynamically adapts model generation timing based on workload execution states. Models are generated when workloads are actively running and performance data is available, rather than using static pre-generated models. This dynamic approach ensures models reflect current workload characteristics while optimizing processing timing.
3Reliability
If model quality evaluation is performed before detection, then false positives are reduced, but computational overhead increases
Solution Approach 1:
The system changes evaluation parameters by assessing model quality metrics such as data coverage, curve fit quality, and statistical significance before deploying models for detection. By evaluating these specific parameters, the system ensures models meet minimum quality thresholds, reducing false positives while avoiding exhaustive computational analysis of all possible model attributes.
Data Source
AI summary
An apparatus comprises a processing device configured to obtain performance data for a plurality of workloads, to select a subset of the performance data corresponding to a subset of the plurality of workloads having a given workload type, and to generate a model characterizing an expected performance of the given workload type by analyzing the selected subset of the performance data to estimate a queuing curve characterizing the expected performance of the given workload type. The processing device is also configured, responsive to determining that a quality of the generated model is above a designated threshold quality level, to utilize the generated model to identify performance impacting events for a given workload of the given workload type and to modify provisioning of compute, storage and network resources allocated to the given workload responsive to identifying performance impacting events for the given workload of the given workload type.