Container Scheduling via Local Storage Cache Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current container scheduling methods fail to ensure successful deployment due to insufficient local storage on compute nodes, even when internal memory is sufficient, as they do not effectively consider local storage resources, leading to scheduling failures, especially on nodes with smaller disc specifications.
Innovation Solution
A method and apparatus that acquire image file information and cache data of compute nodes to determine the local storage resource requirement for each node, identifying candidate nodes with sufficient resources and prioritizing them based on resource availability and cache-hit ratios to select the most suitable node for scheduling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If container scheduling is performed based only on internal memory availability, then memory resources are efficiently utilized, but local storage resources may be exhausted leading to scheduling failures
Solution Approach 1:
The scheduling method performs preliminary checks on both internal memory and local storage resources before allocating containers. By evaluating storage capacity in advance and predicting future storage needs based on cached image layers, the system prevents scheduling failures before they occur, ensuring both memory and storage resources are sufficient for container deployment
Solution Approach 2:
The system implements feedback mechanisms by monitoring local storage usage levels and adjusting scheduling decisions accordingly. When storage resources are approaching exhaustion, the scheduler receives feedback and modifies its behavior to avoid selecting such nodes, thereby maintaining high scheduling success rates while efficiently utilizing available resources
2Productivity
If compute nodes with smaller disc specifications are used for scheduling, then hardware resource utilization is improved, but storage capacity becomes insufficient for caching container images
Solution Approach 1:
The scheduling method applies local quality by evaluating each compute node's specific storage characteristics and matching them with appropriate container workloads. Nodes with smaller storage are strategically selected for containers with smaller image layers, while nodes with larger storage handle more demanding applications, thereby efficiently utilizing hardware resources without exceeding storage capacity
Solution Approach 2:
The system dynamically changes the selection parameters for compute nodes based on current storage conditions and container requirements. By adjusting the weighting and thresholds for storage capacity versus other resources like CPU and memory, the scheduler can adapt to different scenarios, maximizing hardware utilization while ensuring sufficient storage for image caching
Data Source
AI summary
A container application scheduling method and apparatus, an electronic device, and a storage medium. The method includes: acquiring image file information of a container application to be scheduled and image file cache information of all compute nodes; where the image file cache information includes local storage residual resources and cached container image layer file information of the compute nodes; determining a local storage resource requirement of the container application for each compute node according to the image file information and cached container image layer file information of each compute node; and determining, according to the local storage resource requirement and the local storage residual resources of each compute node, a target compute node configured for performing container application scheduling.