Riemannian Workload Scoring for System Configuration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in accurately determining the appropriate infrastructure to support customer workloads, often leading to inefficiencies and increased costs due to inadequate sizing and configuration.
Innovation Solution
The method involves training Riemannian models for each pair of workload classes and system configurations using telemetry data, deploying these models with a new system, and continuously evaluating telemetry data to generate scores, which are used to recommend optimal system configurations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If infrastructure sizing is made a priori without workload execution data, then system configuration can be determined in advance, but the accuracy of matching infrastructure to workload requirements deteriorates
Solution Approach 1:
The system performs preliminary action by training Riemannian models offline using telemetry data from workload executions across multiple system configurations. This pre-computation stores workload characteristics and infrastructure responses in advance, enabling rapid online assessment without executing actual workloads during configuration determination, thus resolving the contradiction between advance sizing and accurate matching.
2Adaptability or versatility
If traditional sizing methods are used without dynamic workload assessment, then infrastructure configuration can be determined statically, but the ability to adapt to workload changes deteriorates
Solution Approach 1:
The patent replaces traditional mechanical/manual infrastructure sizing methods with a mathematical Riemannian geometry-based system. The Riemannian model uses covariance matrices and geometric distances in a high-dimensional space to assess workload characteristics and match them to optimal configurations, substituting complex manual analysis with automated mathematical computation that adapts dynamically to workload changes.
Solution Approach 2:
The system changes parameters by using Riemannian distance metrics and covariance matrix transformations to dynamically assess workload characteristics. Instead of static configuration rules, the system computes distances in a Riemannian manifold space, allowing adaptive parameter adjustment based on actual workload telemetry data, thereby achieving adaptability without proportionally increasing system complexity.
3Measurement precision
If extensive telemetry data collection and model evaluation is performed, then workload characterization accuracy is improved, but computational overhead increases
Solution Approach 1:
The system performs preliminary action by pre-training Riemannian models offline using extensive telemetry data from workload executions. This offline computation phase processes large amounts of data to build standardized models for different workload classes and system configurations. During online operation, only lightweight model evaluation is required, significantly reducing computational overhead while maintaining high characterization accuracy.
Data Source
AI summary
One example method includes deploying a set of non-degenerate models to a system having a known configuration, where each of the non-degenerate models corresponds to a pair that comprises a system configuration and a workload class, running a workload on the system, collecting telemetry data generated as a result of the running of the workload, assessing the telemetry data with each of the non-degenerate models to generate a respective score for each of the models, identifying, as among the non-degenerate models, which of the non-degenerate models has the best score, and determining, based on the best score, whether or not a change is needed to hardware and/or software of the known configuration of the system.