Electronic Request Processing With Containerized Model Isolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing software solutions face challenges such as system crashes due to invalid inputs, resource monopolization leading to slow downs, and inflexibility due to shared third-party dependencies, resulting in inefficient processing of electronic requests.
Innovation Solution
A system architecture that separates the API layer from the model execution layer, validates requests, and executes models in individual containerized environments, allowing independent resource allocation and dependency management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the API layer and model execution layer are executed in the same environment, then the system structure is simple, but the system is prone to crashes from invalid inputs and resource monopolization
Solution Approach 1:
The system divides the execution environment into separate containers for the API layer and model execution layer. Each container operates independently with its own process space, preventing crashes in one layer from affecting the other. This segmentation isolates invalid inputs and resource issues to specific containers while maintaining overall system stability.
2Productivity
If multiple predictive models share the same resource pool, then resource utilization is efficient, but one model can monopolize CPU and cause system slow downs
Solution Approach 1:
Each predictive model is executed in a separate containerized environment with isolated CPU resources. This prevents any single model from monopolizing the CPU, as each container has its own allocated resources. The system maintains efficient resource utilization by allowing multiple containers to run in parallel while ensuring fair resource distribution through container isolation.
3Adaptability or versatility
If the API layer loads models from a cross-platform database, then model storage is flexible, but loading models takes long duration causing timeouts
Solution Approach 1:
The system pre-loads predictive models into the containerized execution environment before requests arrive. Models are prepared and staged in advance within the container, so when a request comes in, the model is already available in memory and can execute immediately without time-consuming loading operations. This preliminary action eliminates timeouts while maintaining the flexibility of cross-platform database storage.
4Device complexity
If all models share the same third-party dependencies, then the dependency management is simple, but upgrading one model breaks functionality of other models
Solution Approach 1:
Each containerized model execution environment has its own isolated set of third-party dependencies. This segmentation allows each model to have its specific dependency versions without affecting other models. When upgrading a model, only its container's dependencies need to be updated, preventing breaking changes for other models while maintaining simple overall dependency management through container isolation.
Data Source
AI summary
Disclosed herein are embodiments of systems, methods, and products comprises a server for efficiently processing electronic requests. The server receives a plurality of predictive computer models and a specification file for each model for registration. The server extracts validation codes for each model based on the specification file. When the server receives an electronic request, the API layer of the server validates the request by verifying the inputs of the request satisfying the validation codes of the corresponding model. If the electronic request is invalid, the server returns an error message immediately; otherwise, the API layer of the server sends the electronic request to the model execution layer. Within the model execution layer, the server executes the corresponding model based on the request inputs and generates output results. The model execution layer transmits the output results back to the API layer, which transmits the output results to the user device.