Dynamic Grid Scheduling Policy for Heterogeneous Resource Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Static schedulers in grid computing systems fail to efficiently utilize resources across heterogeneous environments, leading to inefficiencies in scheduling jobs due to specialized policies and lack of global dynamic scheduling, resulting in underutilization of resources and inability to manage diverse application and resource types.
Innovation Solution
A dynamic grid scheduling policy is created using resource availability information, static scheduling policies, and status from informational providers or agents, optimizing resource allocation across nodes in a heterogeneous grid computing system, allowing for efficient scheduling of grid jobs regardless of resource types and application types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If static scheduling policies are used to manage dedicated resources on defined nodes, then scheduling simplicity is maintained, but resource utilization efficiency deteriorates due to inability to dynamically allocate resources across heterogeneous environments
Solution Approach 1:
The system segments the scheduling function into two parts: a dynamic grid scheduling policy that handles high-level resource allocation decisions, and multiple static schedulers that handle execution on specific nodes. This segmentation allows the dynamic policy to optimize resource utilization while the static schedulers maintain operational simplicity at the execution level.
Solution Approach 2:
The patent introduces a dynamic grid scheduling policy that can adaptively adjust resource allocation based on current system state, job requirements, and resource availability. This dynamic approach replaces rigid static scheduling with flexible decision-making that responds to changing conditions, thereby improving resource utilization efficiency.
2Productivity
If specialized static scheduling policies are implemented for particular application classes, then scheduling effectiveness for those applications improves, but adaptability to diverse application and resource types deteriorates
Solution Approach 1:
The dynamic grid scheduling policy serves as a universal scheduling mechanism that can handle diverse application types and resource configurations. Instead of requiring separate specialized schedulers for each application class, the dynamic policy adapts its behavior based on job characteristics and resource availability, providing both effectiveness and versatility through a single unified approach.
Solution Approach 2:
The system changes scheduling parameters dynamically based on application requirements and resource states. The dynamic grid scheduling policy adjusts allocation strategies, priority levels, and resource selection criteria according to the specific characteristics of each job and the current environment, enabling effective scheduling across diverse application types without requiring specialized policies for each.
3Ease of operation
If static schedulers manage nodes with dedicated resource allocation, then resource management simplicity is maintained, but resource availability for other jobs deteriorates leading to idle nodes
Solution Approach 1:
The dynamic grid scheduling policy acts as an intermediary layer between the static schedulers and the job queue. It monitors resource availability across all nodes and dynamically allocates jobs to appropriate resources, preventing idle nodes while maintaining the simplicity of static scheduler operation. The intermediary coordinates resource sharing without requiring complex changes to the underlying static scheduling infrastructure.
4Adaptability or versatility
If multiple static schedulers with different policies are deployed in heterogeneous grid systems, then support for diverse resource types improves, but global coordination capability deteriorates
Solution Approach 1:
The patent merges the coordination function of multiple static schedulers into a single dynamic grid scheduling policy. This policy consolidates resource information from all static schedulers and makes centralized allocation decisions, thereby achieving global coordination while preserving the ability to support diverse resource types through the adaptive nature of the dynamic policy.
Data Source
AI summary
The illustrative embodiments described herein provide a computer-implemented method, apparatus, and computer program product for scheduling grid jobs. In one embodiment, a process identifies information describing available resources on a set of nodes on a heterogeneous grid computing system to form resource availability information. The process identifies a set of static scheduling policies for a set of static schedulers that manage the set of nodes. The process also identifies a static scheduling status for a portion of the set of static schedulers. The process creates a dynamic grid scheduling policy using the resource availability information, the set of static scheduling policies, and the static scheduling status. The process also schedules a set of grid jobs for execution by the available resources using the dynamic grid scheduling policy.