Synthetic Workload Generation for Migration Impact Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting the impact of workload migration to a new platform are unreliable, especially when confidentiality concerns prevent deploying the workload on a new system, and require intrusive monitoring or significant effort, making it difficult to assess performance changes without compromising security or incurring high costs.
Innovation Solution
Generating a synthetic workload with the same utilization pattern as the original workload, which can be executed on both the original and target platforms to predict performance changes, allowing for non-intrusive monitoring and cost-effective assessment of migration impacts without exposing confidential data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the workload is deployed on a new platform to assess performance impact, then migration performance prediction can be obtained, but confidential data is exposed to the new vendor
Solution Approach 1:
The patent creates a synthetic workload that copies the utilization pattern of the original confidential workload without containing the actual confidential data. This synthetic workload is then executed on the new platform to predict performance impact, thereby obtaining accurate performance measurements while keeping the original confidential data secure and never exposed to the new vendor.
2Object-affected harmful factors
If the workload is not deployed on the new platform due to security concerns, then data confidentiality is maintained, but performance prediction becomes unreliable
Solution Approach 1:
The patent introduces a synthetic workload as an intermediary between the confidential original workload and the new platform. This intermediary preserves the performance characteristics and utilization patterns needed for accurate prediction while eliminating the security risk of exposing actual confidential data during migration assessment.
3Measurement precision
If intrusive monitoring is used to track workload performance, then detailed performance metrics are obtained, but system complexity and deployment difficulty increase
Solution Approach 1:
The patent employs a synthetic workload that is lightweight and purpose-built for performance assessment. This synthetic workload can be easily deployed, executed, and removed without requiring complex permanent monitoring infrastructure, thereby obtaining detailed performance metrics while minimizing system complexity and deployment overhead.
Data Source
AI summary
A method, system and product for predicting impact of workload migration. The method comprising: obtaining a utilization pattern of a workload that is being executed on a first platform; generating a synthetic workload that is configured to have the utilization pattern when executed on the first platform; executing the synthetic workload on a second platform; and identifying a change in performance between execution of the synthetic workload on the first platform and between execution of the synthetic workload on the second platform in order to provide a prediction of an impact of migrating the workload from the first platform to the second platform.