Cloud Storage Volume Scheduling Using Multi-Dimensional Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cloud storage systems suffer from non-uniform resource allocation, affecting the speed of responding to data requests and data transmission due to load balancing being based solely on capacity dimensions, neglecting other factors like storage performance.
Innovation Solution
A cloud storage resource scheduling method that utilizes time-series statistical data to generate a resource scheduling topological graph, selecting candidate storage volumes for scheduling based on multiple resource dimensions, and generating a resource scheduling plan to balance load across storage clusters, thereby optimizing resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If resource allocation is based solely on capacity dimensions, then storage capacity utilization is improved, but resource allocation uniformity deteriorates
Solution Approach 1:
The patent extends the resource allocation model from single-dimension capacity-based scheduling to multi-dimensional scheduling that incorporates both capacity resources and performance resources. This dimensional expansion allows the system to simultaneously optimize storage capacity utilization and resource allocation uniformity by considering multiple factors including IOPS, bandwidth, and latency alongside raw storage capacity.
Solution Approach 2:
The patent introduces dynamic parameter adjustment by using time-series statistical data to capture resource state changes over time. The system monitors and adjusts allocation based on temporal patterns in resource usage, transforming static capacity-based allocation into dynamic multi-parameter optimization that adapts to changing workloads and resource states.
2Device complexity
If single-dimension load balancing is implemented, then implementation complexity is reduced, but response speed and data transmission speed deteriorate
Solution Approach 1:
The patent segments the resource scheduling problem into distinct capacity resource scheduling and performance resource scheduling components. By dividing the multi-dimensional optimization problem into manageable segments that can be addressed separately through topological graph analysis, the system achieves complex multi-dimensional load balancing without proportionally increasing implementation complexity.
Solution Approach 2:
The patent introduces a resource scheduling topological graph as an intermediary data structure that models resource relationships and enables efficient multi-dimensional optimization. This topological representation serves as a mediator between raw resource data and scheduling decisions, allowing the system to handle complex multi-dimensional constraints through graph-based algorithms rather than direct complex optimization computations.
3Ease of operation
If multi-dimensional load balancing is implemented, then resource allocation uniformity is improved, but computational complexity increases
Solution Approach 1:
The patent performs preliminary actions by pre-computing and storing time-series statistical data that captures resource state changes over time. This pre-processing of resource data into structured time-series formats enables the multi-dimensional load balancing algorithm to operate on prepared data structures, reducing the computational burden during actual scheduling operations while maintaining allocation uniformity.
Solution Approach 2:
The patent replaces direct complex multi-dimensional optimization computations with topological graph-based analysis. By substituting traditional numerical optimization methods with graph theory approaches, the system achieves multi-dimensional load balancing through structural analysis of resource relationships rather than computationally intensive iterative optimization, reducing algorithmic complexity while maintaining allocation uniformity.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Embodiments of the present disclosure provide a cloud storage resource scheduling method and apparatus, an electronic device, and a storage medium. The method includes: obtaining (S101) time-series statistical data of a storage system, which is used for representing a change over time of a plurality of resource states of a storage cluster in the storage system; determining (S102), based on the time-series statistical data, whether resource scheduling needs to be triggered; in response to determining resource scheduling needs to be triggered, generating (S103) a resource scheduling topological graph, and selecting, from the storage cluster, candidate storage volume sets, based on the resource scheduling topological graph and the time-series statistical data; generating (S104) a resource scheduling plan based on a resource situation of each candidate storage volume, and controlling (S105) a performing of a scheduling task for a related storage volume based on the resource scheduling plan.