Storage Resource Scheduling by Task Type and Service Level
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In cloud computing, existing storage and computing systems manage resources independently, limiting the ability to schedule and optimize storage resources based on specific task types and service levels, leading to inefficient resource allocation and potential impact on quality of service (QoS).
Innovation Solution
A storage resource scheduling method where the computing system identifies task types and sends corresponding information to the storage system, which acquires and applies scheduling policies to allocate storage units, allowing for dynamic scheduling and management of resources based on task types, service levels, and performance indicators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If storage and computing systems are managed independently with separate resource allocation, then system simplicity and ease of management are maintained, but resource utilization efficiency and QoS optimization deteriorate
Solution Approach 1:
The patent merges storage resource management and computing resource management into a unified system. The storage resource scheduling method integrates with the computing system to enable joint optimization of resource allocation based on task types and service levels, thereby improving overall resource utilization efficiency while maintaining manageable system complexity through modular integration.
Solution Approach 2:
The storage system is designed to perform multiple functions: not only providing storage capacity but also actively participating in resource scheduling decisions based on computing task requirements. The system can dynamically adjust storage resource allocation according to different task types (e.g., backup, replication, primary storage) and service levels, making the storage infrastructure universally adaptable to various computing workloads.
2Adaptability or versatility
If storage resources are allocated at the granularity of storage units only, then storage management simplicity is maintained, but the ability to optimize resources based on specific task requirements deteriorates
Solution Approach 1:
The patent segments the storage resource scheduling into different policy types based on task characteristics: backup policies, replication policies, and primary storage policies. Each policy type addresses specific task requirements with tailored parameters. This segmentation enables task-specific optimization without requiring a completely complex unified scheduling mechanism, as each segment can be independently configured and managed.
Solution Approach 2:
The system applies different scheduling policies and optimization strategies to different storage units based on their specific task assignments. Instead of using a uniform allocation approach, the system tailors resource allocation parameters (such as priority, bandwidth, IOPS) to the local requirements of each task type and service level, thereby achieving task-specific optimization while keeping the overall system manageable through localized policy application.
3Productivity
If SLA levels are configured in advance for each VM, then QoS predictability is improved, but the ability to dynamically adjust resources based on actual task types deteriorates
Solution Approach 1:
The patent introduces dynamic adjustment mechanisms that allow the system to adapt storage resource allocation in real-time based on actual task types and workload characteristics. While SLA levels provide a baseline QoS guarantee, the system can dynamically modify resource allocation within the SLA framework according to the specific task type (backup, replication, primary storage) and current system conditions, thereby improving resource allocation efficiency while maintaining QoS stability through bounded dynamic adjustments.
Solution Approach 2:
The system implements feedback mechanisms where storage resource allocation is continuously monitored and adjusted based on task performance and system state. The scheduling policies incorporate feedback from task execution results and resource utilization metrics to optimize allocation dynamically. This feedback loop ensures that QoS guarantees are maintained while allowing efficient resource utilization adjustments based on actual task requirements and system conditions.
Data Source
AI summary
A storage resource scheduling method and a storage and computing system, where the storage and computing system has a computing system and a storage system, the computing system has at least one computing unit, and the storage system has at least one storage unit. The method executed by the computing system includes: identifying a task type of a computing unit in the at least one computing unit; sending task type information to the storage system, where the task type information carries the task type; acquiring a scheduling policy of the task type according to the task type information; and scheduling, according to the scheduling policy, a storage unit corresponding to the computing unit. In the method, different tasks of a computing unit are perceived, and resource scheduling is performed according to a task type, thereby implementing scheduling and management on different tasks of a same storage unit.