Performance Characterization, Optimization for Leading Company

A leading semiconductor company was looking at optimizing the performance of a key business solution. MiddlewareWorks was engaged to review the application, characterize the workloads and optimize the performance.

The Approach

Engineers from MiddlewareWorks Technology group performed workload analysis, setup automated tests, tuning and optimization, final load scalability and sizing projections design and code from a clusterability point of view, and ran and analyzed the application. The engineers performed a scalability analysis, along with a comprehensive analysis and tuning for the application, the application server, the hardware, and stacks.

Optimization

The Engineers monitored and analyzed the net performance of the system and adjusted its configuration to obtain optimum performance. Some others goals of this performance tuning exercise were to maximize concurrent users, throughput, and reliability. In other words, the focus was on the ability to service more users faster and not break down in the process.

The MW "Fahrenheit" Process

The following process was used for analyzing performance and testing:

All the performance tests were based on expected usage patterns and workloads.

Key Findings

The response time, that is time taken to process each request generated, was low, and there was network latency/delay in sending the data back and forth. The total number of concurrent users was high, and the throughput was not optimum.

The key bottlenecks were fixed and further observations were made to better the architecture and performance of the business solution.