Fused Container Deployment for Cascading FaaS Functions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
FaaS deployment faces challenges with cascading execution of functions leading to increased resource consumption, execution time, and low performance due to serial provisioning of functions, which can result in higher costs and inefficient resource utilization.
Innovation Solution
A system dynamically collects information on deployed functions and fuses them into a single container, using static or dynamic analysis to identify cascading function calls and optimize resource usage, reducing compute resources and improving performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If functions are deployed as separate containerized functions with cascading execution, then function modularity and independence are improved, but resource consumption and execution time increase
Solution Approach 1:
The patent merges multiple containerized functions that execute in cascade into a single fused container. The system identifies cascading function calls, packages their source code together into one image, and deploys them as a unified entity. This combining approach maintains the functional modularity benefits while reducing resource overhead by eliminating redundant container infrastructure for each function.
Solution Approach 2:
The fused container becomes a universal execution unit that can handle multiple function invocations internally. Instead of requiring separate container instances for each function call in the cascade, a single fused container handles all functions, making the container infrastructure universal and reducing overall resource consumption while preserving function independence through internal code organization.
2Adaptability or versatility
If functions are deployed as separate containerized functions with cascading execution, then function independence is improved, but execution time increases
Solution Approach 1:
By merging cascading functions into a single container, the system eliminates the time required for inter-container communication and sequential provisioning. The fused container executes functions in a unified process space, removing overhead associated with container lifecycle management between function calls while preserving logical function independence through code structure.
3Adaptability or versatility
If functions are provisioned serially in cascade, then function deployment flexibility is improved, but productivity decreases
Solution Approach 1:
The system combines serially-provisioned functions into a parallel-execution unified container. The fused container can execute multiple functions concurrently or in optimized sequences within a single process space, dramatically improving productivity by eliminating the serial provisioning bottleneck while maintaining deployment flexibility through the containerization framework.
4Reliability
If multiple separate container instances are used for cascading functions, then function isolation is improved, but device complexity increases
Solution Approach 1:
The patent reduces system complexity by merging multiple container instances into one fused container. This consolidation simplifies the deployment architecture, reduces the number of managed entities, and lowers operational complexity while maintaining function isolation through internal code boundaries and execution contexts within the unified container.
Data Source
AI summary
A system can determine that a first containerized function invokes a second containerized function. The system can access first source code of the first containerized function from a repository. The system can access second source code of the second containerized function from the repository. The system can package the first source code and the second source code into an image that comprises a container in which the first source code and the second source code are configured to execute. The system can deploy the image to produce a deployed image. The system can terminate a first executing instance of the first containerized function. The system can terminate a second executing instance of the second containerized function. The system can direct a first call to invoke the first containerized function to the deployed image. The system can direct a second call to invoke the second containerized function to the deployed image.


