Server Bottleneck Detection via Execution History Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods lack a systematic way to detect and address bottlenecks in server performance during concurrent transaction processing, requiring expert intervention and significant time, which can lead to inefficient system performance and increased costs.
Innovation Solution
A bottleneck detection system comprising a tracer, analyzer, and determination section that collects and analyzes execution histories to measure performance indices, using load generators and probes to identify bottlenecks by comparing resource utilization and elapsed times under varying loads, and applying algorithms to determine the cause of bottlenecks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If performance evaluation models are used to evaluate server performance, then evaluation speed is improved, but measurement precision deteriorates because bottlenecks not reflected in the model are not detected
Solution Approach 1:
The system segments bottleneck detection into two parts: model-based evaluation for speed and trace-based actual measurement for precision. The trace data is divided into multiple records with timestamps, allowing detailed analysis of execution flow without slowing down the overall evaluation process.
Solution Approach 2:
The invention introduces trace data as an intermediary between the performance evaluation model and actual server performance. The trace data captures actual execution flow and resource usage, serving as a mediator to identify bottlenecks that the model cannot detect, thereby improving measurement precision without sacrificing evaluation speed.
2Measurement precision
If expert intervention is used to detect bottlenecks, then measurement precision is improved, but loss of time increases significantly
Solution Approach 1:
The system enables self-service bottleneck detection by automatically analyzing trace data through the determination section. The server itself generates and provides the trace data, and the system automatically processes this data to identify bottlenecks without requiring external expert intervention, thereby eliminating time loss while maintaining high measurement precision.
Solution Approach 2:
The system implements feedback by continuously collecting trace data during server operation and automatically analyzing it to provide bottleneck information. This closed-loop feedback mechanism replaces manual expert analysis with automated real-time or near-real-time bottleneck detection, significantly reducing the time required while maintaining accuracy.
3Measurement precision
If multiple measurement tools are combined to detect bottlenecks, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The invention merges the bottleneck detection functionality into the server software itself through integrated trace mechanisms. Instead of combining multiple external measurement tools, the trace data collection and analysis functions are merged with the server's transaction processing, creating a unified system that reduces complexity while maintaining measurement precision.
Solution Approach 2:
The trace mechanism serves multiple functions: it records execution flow, captures resource usage, provides timing information, and enables bottleneck analysis all through a single integrated system. This multi-functional approach replaces the need for multiple separate measurement tools, reducing device complexity while maintaining comprehensive bottleneck detection capability.
Data Source
AI summary
A bottleneck detection system, a measurement object server, a bottleneck detection method and a program capable of specifying a factor that causes a bottleneck on the occasion of concurrent processing of a plurality of transactions. A tracer collects execution histories of a measurement object server processing transactions with a specified amount of load. An analyzer receives the execution histories from the tracer and analyzes them to measure performance indices with respect to software components in the measurement object server. A determination section receives the analysis results from the analyzer and, processes a value that indicates the relation between the amount of load and each of the performance indices by a prescribed algorithm.