Container Image Sharing Across Runtimes to Cut Duplicate Storage

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The challenge of managing multiple container runtime images for the same application leads to excessive storage and bandwidth usage due to the need to download and store duplicate content data for each runtime, which is inefficient and costly.

Innovation Solution

Implementing an image sharing file system that separates and stores content data once across multiple container runtimes, allowing reuse and reducing the need for duplicate storage, and enabling image creation based on stored content data without full repository pulls.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If multiple container runtime images are downloaded and stored separately for the same application, then each runtime can operate independently, but storage space and bandwidth consumption increase significantly

Engineering Contradiction:

Improveruntime independenceVSAvoidstorage space

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The container image is segmented into two distinct parts: content data (application files, dependencies, environment) and management data (runtime-specific configuration, metadata). This segmentation allows the content data to be shared across multiple runtimes while management data remains runtime-specific, reducing overall storage requirements while maintaining runtime independence.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The content data portion of the image is designed to be universal and shareable across different container runtimes (Docker, LXC, Singularity). By making the content data multi-functional and runtime-agnostic, the same content data can serve multiple runtime environments, eliminating the need to store duplicate content data for each runtime while still allowing each runtime to operate independently with its own management data.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Reliability

If multiple container runtime images are downloaded from the repository, then each runtime has its complete image, but communication bandwidth usage increases

Engineering Contradiction:

Improveimage completenessVSAvoidbandwidth consumption

Core Design Contradiction:

ReliabilityVSLoss of energy

Solution Approach 1:

The management data is extracted and separated from the content data. This extraction allows the content data to be obtained once and shared across multiple runtimes, while only the smaller management data portions need to be downloaded separately for each runtime, significantly reducing total bandwidth consumption while maintaining image completeness for each runtime.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The content data is downloaded and stored in advance in a shared location accessible by multiple runtimes. This preliminary action eliminates the need to re-download content data for subsequent runtimes, reducing bandwidth consumption. Only the runtime-specific management data needs to be downloaded separately, making the overall process more efficient.

Inventive Principle:

Principle #10Preliminary action

3Ease of operation

If content data is stored separately for each container runtime, then each runtime has dedicated data, but storage efficiency decreases

Engineering Contradiction:

Improvedata managementVSAvoidstorage efficiency

Core Design Contradiction:

Ease of operationVSQuantity of substance

Solution Approach 1:

The content data from multiple container runtime images is merged into a single shared storage location. This merging eliminates duplicate storage of identical content data across different runtimes, significantly improving storage efficiency. Each runtime maintains its own management data, ensuring ease of operation and data management while benefiting from the shared content data repository.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS12487844B2Image management for container runtimes

Publication Date: 2025.12.02 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

Disclosed is an approach for image management for container runtimes. A first image of a first application for a first container runtime can be downloaded from an image repository providing the first image. Content data of the first application and management data corresponding to the first application and the first container runtime can be extracted from the first image. Both the content data of the first application and the management data corresponding to the first application and the first container runtime can be stored in an image sharing file system in response to the content data of the first application being not stored in the image sharing file system. Only the management data corresponding to the first application and the first container runtime can be stored in the image sharing file system in response to the content data of the first application being stored in the image sharing file system.