Distributed Application Interface Performance Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current monitoring tools are inadequate for continuously monitoring interface performance in distributed application systems, particularly failing to effectively analyze and manage the tantile of interface response times across large-scale, complex architectures.
Innovation Solution
A computer-implemented method and apparatus that intercepts interface calls to obtain and update performance data distribution characteristics, using algorithms like T-Digest or histograms to estimate tantiles, and integrates with other monitoring tools for comprehensive, real-time performance analysis across distributed systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If stress testing tools like JMeter or LoadRunner are used to monitor interface performance, then interface response times can be recorded and averaged, but the tools are limited to stress testing phase and cannot continuously monitor online systems
Solution Approach 1:
The monitoring system transitions from static stress testing to dynamic continuous monitoring by implementing real-time data collection from distributed application instances. The system dynamically adapts to online environments by continuously gathering performance metrics without interrupting service operations.
Solution Approach 2:
The monitoring tool is designed to serve multiple functions: stress testing, continuous online monitoring, and distributed system observation. By making the tool universal, it can operate across different phases (testing and production) and architectures (monolithic and distributed), eliminating the limitation of being confined to stress testing only.
2Measurement precision
If Prometheus or other monitoring tools are used to monitor interface performance, then interface response times can be monitored, but the tools are only suitable for single instance and simple system architectures like monolithic or active-standby
Solution Approach 1:
The monitoring system is segmented into multiple independent agents deployed on different instances of the distributed application. Each agent collects local performance metrics independently, and the results are aggregated to form a comprehensive view of the distributed system's performance.
Solution Approach 2:
The monitoring capability extends from single-instance to multi-instance by adding the dimension of distribution. The system monitors across multiple spatial dimensions (different instances, nodes, or services) simultaneously, enabling comprehensive monitoring of distributed architectures.
3Area of stationary object
If traditional monitoring tools are applied to distributed application clusters, then monitoring coverage can be expanded, but it becomes difficult to effectively analyze and manage tantile of interface response times across large-scale complex architectures
Solution Approach 1:
An intermediary aggregation layer is introduced between the distributed instances and the analysis system. This intermediary collects, standardizes, and pre-processes performance data from multiple instances, reducing the complexity of data analysis by providing a unified view of distributed system performance.
Solution Approach 2:
The system creates simplified representations (copies) of the complex distributed system's performance characteristics. By modeling and aggregating data into standardized formats, the system enables effective analysis without being overwhelmed by the raw complexity of distributed architectures.
4Productivity
If comprehensive monitoring of distributed systems is implemented, then real-time performance analysis can be achieved, but computational resources and development costs increase
Solution Approach 1:
The monitoring system implements partial monitoring by focusing on key performance indicators and critical interfaces rather than attempting to monitor every aspect of the distributed system. This selective approach achieves effective real-time analysis while minimizing computational overhead and resource consumption.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
The present disclosure discloses a method and apparatus of monitoring an interface performance of a distributed application, a device and a storage medium, which relates to a field of computer technology, in particular to a field of cloud platform. The method includes: in case of detecting a caller request for calling an interface of the distributed application, obtaining a performance data of the interface for responding the caller request; updating a performance data distribution characteristic of the interface according to the performance data of the interface for responding the caller request, so as to obtain an updated performance data distribution characteristic; and monitoring the interface performance of the distributed application, according to the updated performance data distribution characteristic of the interface.