Event-Driven Transaction Performance Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current performance testing tools face challenges in measuring application performance, especially in complex Web 2.0 applications with asynchronous communications, due to limitations in script-based transaction models that struggle with external dependencies and mismatched measurement levels, making it difficult to accurately evaluate performance across all system components.
Innovation Solution
Implementing an event-driven technique that generates transactions independently of scripts, using predefined rules to define transactions in a declarative manner, allowing for synchronous or asynchronous event dispatching and transformation into transactions, enabling performance measurement across various system components without script dependency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If script-based transaction models are used for performance testing, then transaction measurement can be implemented, but measurement accuracy deteriorates due to external dependencies and mismatched measurement levels
Solution Approach 1:
The patent extracts the transaction measurement logic from complex performance testing scripts into a separate event-driven transaction model. This allows transactions to be defined independently based on event sequences, eliminating the need for scripts to handle both user interaction simulation and performance measurement, thereby improving measurement accuracy without increasing overall system complexity
Solution Approach 2:
The patent segments the performance testing system into distinct components: event generation, event processing, and transaction measurement. By separating transaction definition from script execution, each component can be optimized independently, with transactions being measured at the appropriate granularity level without being constrained by script structure
2Productivity
If synchronous event processing is used, then transaction measurement is simpler, but performance deteriorates due to blocking operations in asynchronous systems
Solution Approach 1:
The patent introduces an event queue as an intermediary between event generation and event processing. Events are placed in the queue asynchronously without blocking the generating thread, while processing occurs independently at appropriate times. This mediator mechanism allows the system to maintain high productivity while accurately measuring transaction times without blocking operations
Solution Approach 2:
The patent records event timestamps at the moment of event occurrence without immediate processing. Event data is captured and stored in the queue with precise timing information, allowing transaction measurement to proceed without waiting for full event processing completion, thus eliminating blocking delays while maintaining measurement accuracy
3Measurement precision
If detailed event tracking is implemented, then measurement precision improves, but system overhead increases
Solution Approach 1:
The patent applies detailed event tracking selectively to specific event types and critical transaction paths rather than uniformly across all system operations. By identifying and monitoring only the events that contribute to performance measurement requirements, the system achieves high measurement precision in critical areas while minimizing computational overhead in less critical areas
Data Source
AI summary
An event-driven performance measurement technique provides a set of transaction definitions that specify transaction actions that correspond to start and end points of a transaction. Events associated with execution of an application are received by a processor-based machine. The machine uses the events to generate transactions in accordance with the set of transaction definitions. The transactions can be used to evaluate performance of the application.