User-Resource Allocation Clustering for Server Access Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for optimizing user-resource allocations in remote-hosted systems fail to accurately quantify multivariate access patterns, leading to excessive allocation errors and potential Service Level Agreement (SLA) violations, especially in systems with high-cardinality userships and resource sets, and lack the ability to dynamically adjust allocations to meet evolving conditions.
Innovation Solution
A system and method that identifies recurrent user-resource access patterns by generating clusters based on historical time-series data, optimizing allocations by determining statistical significance and associating users and resources with servers, using techniques like k-means clustering and statistical p-value calculations to ensure efficient and flexible resource distribution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional partitioning methods are used to allocate users and resources to servers, then the system structure is simple and easy to implement, but allocation errors increase and SLA compliance deteriorates under high-cardinality userships and resource sets
Solution Approach 1:
The patent segments users and resources into distinct clusters based on access patterns. Users are grouped into user clusters and resources into resource clusters, allowing the system to handle high-cardinality userships and resource sets by breaking them down into manageable segments that can be efficiently allocated to servers while maintaining allocation accuracy
Solution Approach 2:
The patent introduces a new dimension of analysis by examining multivariate access patterns across multiple parameters simultaneously. Instead of traditional single-parameter partitioning, the system analyzes correlations between multiple user-resource attributes, transforming the allocation problem from simple partitioning to multidimensional pattern recognition, which improves allocation precision without excessive complexity
2Adaptability or versatility
If static allocation methods are used, then the system is stable and easy to manage, but the system cannot adapt to evolving access patterns and SLA requirements
Solution Approach 1:
The patent implements dynamic allocation by continuously monitoring access patterns and recalculating user and resource clusters based on evolving conditions. The system adapts to changing SLA requirements and access behaviors by periodically updating cluster assignments and re-allocating users and resources to appropriate servers, ensuring the allocation remains optimized as system conditions change
Solution Approach 2:
The system incorporates feedback mechanisms by monitoring SLA compliance and access pattern changes, then using this information to adjust allocations. The patent evaluates allocation performance and feeds this information back into the clustering and allocation process, allowing the system to learn from past performance and continuously improve its adaptability to evolving requirements
3Measurement precision
If detailed monitoring and analysis of access patterns is implemented, then allocation precision improves, but computational resources and processing time increase
Solution Approach 1:
The patent extracts only the essential and most relevant features from access pattern data that are necessary for accurate clustering and allocation. Instead of processing all available data with equal weight, the system identifies and focuses on key pattern characteristics, reducing computational overhead while maintaining the precision needed for accurate user-resource allocation decisions
Data Source
AI summary
Systems, methods, and computer-readable media are provided for facilitating system optimization through the use of user-resource allocations to servers based on determined access patterns. In one embodiment, recurrent patterns of access are identified based on combinations of computer system users and system-hosted resources. In some embodiments, groupings of user-resource combinations can be determined. The groupings are valuable for optimizing the allocation of users and/or resources to a plurality of servers, particularly under conditions of heavy simultaneous resource demand. Patterns may be determined from user-resource pair access time series, and groupings may be determined based on derived strength of association of these. Based on the groupings, users and resources may be allocated to servers efficiently. Allocation optimization can be an effective means for mitigating or preventing Service Level Agreement non-compliance.