Container Image Pre-deployment via Policy-Based Caching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The efficiency of container startup is reduced due to the need to obtain an image from a repository before deploying a container, which consumes time and slows down the deployment process.
Innovation Solution
An image deployment method and apparatus that determine a to-be-deployed image based on configuration policies, such as autoscaling or replication control policies, and popularity information, allowing for pre-deployment of images on nodes with the lowest load, thereby avoiding the need to retrieve images from a repository during container startup.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If images are obtained from repository before deploying containers, then container deployment can be performed, but container startup speed is reduced due to time consumption in obtaining images
Solution Approach 1:
The patent applies preliminary action by pre-downloading and caching images on nodes before container deployment is needed. The image management module proactively obtains images from the repository and stores them in the local image cache, so that when containers need to be deployed, the images are already available locally, eliminating the time-consuming network retrieval step during container startup.
2Productivity
If images are pre-deployed on nodes, then container startup speed is improved, but node resource consumption increases due to storing images
Solution Approach 1:
The patent applies local quality by implementing selective image caching based on node characteristics and deployment needs. Not all nodes cache all images; instead, the system determines which nodes should cache which images based on their resource capabilities, deployment frequency, and availability. This localized approach ensures that images are cached where they are most needed while minimizing overall resource consumption.
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting image caching strategies based on various parameters including node resource status, image size, deployment frequency, and time-to-live. The system can modify caching policies by changing these parameters to balance between startup speed improvement and resource consumption, allowing flexible adaptation to different operational conditions.
3Loss of time
If images are cached on nodes, then image retrieval time is reduced, but image management complexity increases
Solution Approach 1:
The patent applies self-service by implementing automatic image management where the image management module autonomously handles image caching, validation, and removal without requiring manual intervention. The system automatically determines which images to cache, validates their integrity, and manages their lifecycle, reducing the operational complexity of image management while maintaining the benefits of local caching.
Data Source
AI summary
An image deployment method includes obtaining a configuration policy for a container, where the configuration policy for the container includes an autoscaling policy or a replication control policy. The autoscaling policy includes a first image identifier, a maximum quantity of containers corresponding to the first image identifier that can be deployed, and a quantity of currently deployed containers corresponding to the first image identifier. The replication control policy includes a second image identifier and a quantity of containers corresponding to the second image identifier that are expected to be deployed. The method determines an identifier of a to-be-deployed image according to the configuration policy for the container.