Deployment Clustering for Testing Mock Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The complexity of computing systems, such as network-attached storage systems, makes exhaustive testing infeasible due to their high variability in size, software features, and configurations, rendering comprehensive testing impractical within a reasonable timeframe.
Innovation Solution
A system and method that filter, cluster, and select computing system deployments based on relevance and properties for testing, using a filtering component to identify relevant deployments, a clustering component to group them, and a selection component to designate deployments for testing, thereby reducing the time and resources required for testing while ensuring representativeness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If exhaustive testing is performed on all computing system deployments, then testing completeness is improved, but testing time and resources become prohibitively large
Solution Approach 1:
The patent segments the large set of computing system deployments into multiple clusters based on similarity metrics. Instead of testing all deployments individually, the system divides them into manageable groups where representative deployments from each cluster are selected for testing, thereby reducing the overall testing scope while maintaining coverage of diverse system configurations.
Solution Approach 2:
The patent creates mock computing systems that replicate the configurations and properties of selected real deployments. These mock systems serve as test surrogates, allowing testing to be performed on simplified copies rather than all actual systems, thus reducing testing resources and time while preserving the ability to detect configuration-related defects.
2Productivity
If the number of deployments to be tested is reduced, then testing time and resources are reduced, but testing representativeness may deteriorate
Solution Approach 1:
The patent applies local quality by ensuring that each cluster contains deployments with similar local characteristics (configuration properties, scale, architecture), and that the selected representative deployment from each cluster captures those specific local qualities. This approach maintains testing representativeness by preserving the unique characteristics of different deployment categories while reducing the overall number of test cases.
Solution Approach 2:
The patent uses parameter-based clustering where deployments are grouped according to their configuration parameters and properties. By selecting representatives based on parameter diversity across clusters, the system ensures that the reduced test set still covers the full range of parameter variations present in the complete deployment population, thereby maintaining representativeness.
Data Source
AI summary
Systems and methods facilitating selection of computer system deployments to mock for testing are described herein. A method as described herein can include filtering, by a first system operatively coupled to a processor, data relating to deployments of respective second systems as collected from the respective second systems, wherein the filtering is based on relevancy of the deployments to a testing criterion and results in a filtered group of deployments; grouping, by the first system, respective ones of the filtered group of deployments into respective clusters according to a difference function, the difference function being defined according to respective properties of the respective ones of the filtered group of deployments; and designating, by the first system, selected ones of the filtered group of deployments from respective ones of the clusters for testing according to a selection criterion.