Cloud Service Dispatching Module for Online Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for evaluating cloud-based services often fail to accurately predict performance in online environments, as offline tests cannot replicate user responses, leading to discrepancies between offline and online performance.
Innovation Solution
An online cloud-based service processing system that includes application service units and evaluation units, with a dispatching module to split service requests between regular and evaluation functions, allowing real-time evaluation and ranking of application services based on output results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If offline tests are used to evaluate cloud-based services, then service performance can be examined using history data, but the tests cannot instantly acquire response from users and performance may differ during online service provision
Solution Approach 1:
The system segments service requests into two distinct paths: a first path for production services and a second path for evaluation services. This segmentation allows offline evaluation units to process real user requests separately from production units, enabling accurate performance measurement without affecting service reliability. The dispatching module routes requests to appropriate units based on their function.
Solution Approach 2:
The patent introduces an intermediary evaluation management module that coordinates between production services and evaluation services. This intermediary manages the registration, activation, and operation of offline evaluation units, ensuring they receive appropriate traffic while maintaining separation from production workflows. The intermediary enables accurate evaluation by mediating the flow of real user requests to evaluation units.
2Measurement precision
If real user traffic is used for evaluation, then accurate online performance can be measured, but service availability may be affected
Solution Approach 1:
The system applies partial action by routing only a portion of user requests to evaluation units while the majority continue to production units. The dispatching module implements this by using hash functions to determine request routing, ensuring evaluation units receive sufficient traffic for accurate measurement without compromising overall service availability. This partial routing enables evaluation while maintaining productivity.
Solution Approach 2:
Different parts of the system have different functions: production units handle service delivery while evaluation units handle performance measurement. The dispatching module applies local quality by routing requests based on their intended destination, ensuring production units maintain high availability for users while evaluation units accurately measure performance using real traffic patterns.
3Adaptability or versatility
If multiple evaluation units are deployed to evaluate different services, then comprehensive service assessment is possible, but system complexity increases
Solution Approach 1:
The evaluation management module provides universal functionality by managing multiple offline evaluation units through a single interface. It handles registration, activation, and traffic routing for all evaluation units, reducing system complexity despite supporting multiple services. The module acts as a universal controller that simplifies the management of diverse evaluation units.
Solution Approach 2:
The system segments evaluation functionality into independent offline evaluation units, each capable of evaluating specific services. This segmentation allows comprehensive service assessment while maintaining manageable complexity, as each unit operates independently under the coordination of the evaluation management module. The modular structure enables versatile evaluation without creating monolithic complexity.
Data Source
AI summary
An online cloud-based service processing system, an online evaluation method and a computer program product thereof are provided. The online evaluation method includes: running at least one application service unit of the processing system; executing at least one evaluation registration procedure for at least one application service evaluation unit of the processing system in response to at least one registration request; running the at least one application service evaluation unit; dispatching at least one online service request respectively to the at least one application service unit and the at least one application service evaluation unit, and generating at least one output result corresponding to the at least one online service request; and calculating at least one ranking for each of the at least one application service evaluation unit based on output result(s) generated by the at least one application service evaluation unit among the at least one output result.


