ML Microservice Routing with Nested Headers to Cut API Calls
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional microservice architectures face inefficiencies such as increased architectural footprint, excessive server costs, and inflexible network resource usage due to hardcoded routing and duplicate microservice deployments, especially in cloud environments.
Innovation Solution
Utilizing a machine-learning model trained on prior data payloads and feedback to predict optimal paths through microservices, modifying data payloads with headers to route them efficiently through necessary microservices, thereby avoiding duplicate instances and reducing unnecessary API calls.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple chains of microservices are loaded in parallel to process multiple records, then processing capability is improved, but architectural footprint and server costs increase
Solution Approach 1:
The patent merges multiple microservice chains into a single shared architecture where microservices are deployed once and shared across multiple data payload processing chains. The orchestrator service manages routing to these shared microservices, eliminating the need to load duplicate instances for each record while maintaining parallel processing capability through efficient resource sharing.
Solution Approach 2:
The patent implements universal microservices that can handle multiple types of data payloads and processing tasks through a single deployment. The microservices are designed with multi-functionality, allowing them to serve different processing chains without requiring separate instances, thereby reducing architectural footprint while maintaining processing versatility.
2Ease of operation
If hardcoded routing is used to route data payloads through microservices, then routing path is determined, but flexibility and adaptability are reduced
Solution Approach 1:
The patent replaces static hardcoded routing with dynamic routing that adapts based on the content and requirements of each data payload. The orchestrator service analyzes incoming data payloads and dynamically determines the appropriate microservice chain to use, allowing the system to flexibly respond to different processing needs without requiring pre-defined rigid paths for all scenarios.
Solution Approach 2:
The patent implements feedback mechanisms where the orchestrator service learns from processing outcomes and adjusts routing decisions accordingly. By monitoring which microservice paths successfully process different data payload types, the system adapts its routing strategy over time, improving both efficiency and flexibility while maintaining ease of operation through automated decision-making.
3Reliability
If every record is processed by every microservice on the chain, then completeness of processing is ensured, but network usage and processing time increase
Solution Approach 1:
The patent extracts and applies filtering logic at the orchestrator service level to identify and exclude microservices that are not required for processing specific data payload types. Instead of routing every record through every microservice, the orchestrator analyzes the payload and extracts only the necessary processing steps, thereby maintaining processing completeness for required services while eliminating unnecessary network calls and reducing overall processing time.
4Productivity
If duplicate instances of common microservices are deployed, then processing parallelism is achieved, but server costs and maintenance overhead increase
Solution Approach 1:
The patent merges duplicate microservice instances into a single shared pool of microservices that can be dynamically allocated to different processing chains. The orchestrator service manages the sharing and allocation of these microservices across multiple concurrent processing tasks, achieving parallelism through efficient resource scheduling rather than through duplicate deployments, thereby reducing server costs and maintenance overhead while maintaining processing capacity.
Data Source
AI summary
The method may comprise: receiving a data payload and first data; predicting, by a trained machine learning model based on the first data, a path through a plurality of microservices associated with the data payload; generating a modified data payload by modifying, via the orchestrator service, the data payload to include: a first header, wherein the first header comprises a first microservice destination address associated with a first microservice of the plurality of microservices and a second header nested within the first header, wherein the second header comprises a second microservice destination address associated with a second microservice of the plurality of microservices; forwarding the modified data payload to the first microservice based on the first header for processing; and forwarding the modified data payload to the second microservice based on the second header for processing.