Cloud Storage Scheduling by Traffic Characteristics and Cluster Balance

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing resource balanced scheduling algorithms in cloud storage systems fail to effectively improve performance balance between clusters due to neglecting traffic characteristics of cloud disks and business models, leading to resource imbalances and potential overload.

Innovation Solution

A resource scheduling method that matches scheduling strategies with traffic characteristics of cloud disks and candidate storage clusters, using strategies based on complementary performance, service type, and cloud disk life cycle to optimize cluster allocation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If traditional resource balanced scheduling algorithm is used, then scheduling simplicity is maintained, but performance balance degree between clusters deteriorates

Engineering Contradiction:

Improvescheduling algorithm complexityVSAvoidperformance balance degree

Core Design Contradiction:

Device complexityVSManufacturing precision

Solution Approach 1:

The patent applies parameter changes by introducing traffic characteristics (bandwidth, IOPS, latency) as new parameters for scheduling decisions. The system dynamically adjusts scheduling based on these parameters, matching cloud disks with storage clusters that have complementary traffic patterns, thereby improving performance balance without requiring complex structural changes to the scheduling framework

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent implements dynamics by making the scheduling algorithm adaptive to changing traffic conditions. The system continuously monitors traffic characteristics and dynamically adjusts scheduling decisions, allowing the scheduler to respond to varying workloads and maintain optimal performance balance across clusters as conditions change

Inventive Principle:

Principle #15Dynamics

2Speed

If resource margin-based scheduling is used, then initial scheduling speed is maintained, but resource overload problem worsens

Engineering Contradiction:

Improvescheduling speedVSAvoidresource overload prevention

Core Design Contradiction:

SpeedVSReliability

Solution Approach 1:

The patent applies preliminary action by pre-calculating and storing traffic characteristics for both cloud disks and storage clusters before scheduling occurs. This preparation work is done in advance, allowing the actual scheduling decision to be made quickly by matching pre-computed characteristics, thus maintaining fast scheduling speed while preventing resource overload through informed decisions

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent implements feedback by continuously monitoring the performance and traffic characteristics of storage clusters after scheduling decisions are made. This feedback mechanism allows the system to learn from past scheduling outcomes and adjust future scheduling decisions, preventing resource overload by identifying and avoiding clusters that are approaching capacity limits

Inventive Principle:

Principle #23Feedback

3Quantity of substance

If capacity algorithm is used for secondary scheduling, then capacity overload is reduced, but performance balance improvement deteriorates

Engineering Contradiction:

Improvestorage capacity utilizationVSAvoidperformance balance degree

Core Design Contradiction:

Quantity of substanceVSManufacturing precision

Solution Approach 1:

The patent extends parameter changes by incorporating multiple performance dimensions beyond just capacity, including bandwidth, IOPS, and latency characteristics. The scheduling system evaluates and matches these diverse parameters to achieve both capacity utilization and performance balance, overcoming the limitation of capacity-only algorithms

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS20250370814A1Resource scheduling method based on cloud storage system, electronic device, and storage medium

Publication Date: 2025.12.04 BEIJING VOLCANO ENGINE TECH CO LTD

AI summary

A resource scheduling method based on a cloud storage system, an electronic device and a storage medium are provided. The method includes: obtaining a traffic characteristic of a target cloud disk and a traffic characteristic of each of a plurality of candidate storage clusters in response to a scheduling instruction of the target cloud disk; matching a target scheduling strategy from a preset scheduling strategy set according to the traffic characteristic of the target cloud disk and the traffic characteristic of each of the plurality of candidate storage clusters; in which the scheduling strategy set is used for maintaining a plurality of scheduling strategies configured based on different traffic characteristics; matching at least one candidate storage cluster from the plurality of candidate storage clusters as a target storage cluster according to the target scheduling strategy, and controlling the target cloud disk to be scheduled to the target storage cluster.