Resource Manager Decouples Workload Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In distributed computing environments, the tight coupling of resource management and workload management functions prevents applications from being aware of each other's resource demands, leading to inefficient resource allocation and conflicting policies, resulting in computational complexity and inefficiencies.
Innovation Solution
A common resource manager decouples resource management from workload management, utilizing a hierarchical consumer tree to apply resource allocation policies and dynamically provision resources, allowing multiple applications to share pooled resources and adjust resource availability based on demand.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If resource management functions are embedded with application workload management functions, then workload management is simplified, but resource allocation efficiency deteriorates and applications cannot be aware of each other's resource demands
Solution Approach 1:
The system segments resource management from workload management by introducing a separate resource manager that operates independently of application workload managers. This segmentation allows resource management to be optimized globally while workload managers focus on application-specific tasks, resolving the contradiction between operational simplicity and allocation efficiency.
Solution Approach 2:
The resource manager acts as an intermediary between workload managers and the resource pool. It receives resource requests from workload managers, applies allocation policies, and distributes resources accordingly. This intermediary role enables efficient resource allocation while keeping workload management simple and application-awareness optional.
2Adaptability or versatility
If multiple disjointed resource allocation policies are specified at different levels, then policy flexibility is improved, but computational complexity increases and conflicting behavior occurs
Solution Approach 1:
The system merges multiple disjointed resource allocation policies into a unified hierarchical policy framework. Policies at different levels (organization, department, project, user) are combined and resolved by the resource manager through a consistent evaluation process, eliminating conflicts and reducing computational complexity while maintaining flexibility.
Solution Approach 2:
The patent introduces a hierarchical dimension to policy management, organizing policies across multiple levels of the organization structure. This dimensional approach allows flexible policy specification at various levels while providing a systematic method for resolving conflicts and determining priority, thereby reducing computational complexity.
3Ease of manufacture
If resources are tightly coupled with workload management, then resource allocation is straightforward, but system availability decreases and resource sharing between applications is limited
Solution Approach 1:
By segmenting resource management from workload management, the system achieves higher availability. The independent resource manager can continue operating and allocating resources even when specific workload managers are down, while maintaining straightforward allocation processes through standardized interfaces and policies.
Data Source
AI summary
A technique for managing pooled resources in a distributed computing environment includes a resource manager that receives resource allocation requests from workload managers for running one or more applications. The applications are initiated by a consumer through at least one of the workload managers where each workload manager is configured to translate demand requests from the applications into resource allocation requests. A determination is made whether there are available resources to satisfy the resource allocation requests. Responsive to determining that there is an unmet resource allocation request, a candidate resource meeting some but not all attributes needed to satisfy the unmet resource allocation request is identified and reconfigured to have the attributes needed to satisfy the unmet resource allocation request.