Home

News

Performance Investigation

Overview
Prediction
Optimisation
Testing

Custom Software

Web Site
Development

Case Studies

Links

Contact Details

Predict the performance of a system, before or during a build ...

Introduction | What is involved?

Introduction

Performance prediction allows you to identify, in advance, problem areas of system performance and which improvements can be made to give the best performance gains.

You can also save time and money by focusing expensive person power on the critical parts of a system build.

An independent performance prediction offers many benefits, including:

• reduced risk of performance failure such as system slowdown/lockup
• identification of no hope scenarios; vital for cutting edge and risky projects
• an opportunity to deliver a scalable system with only the currently needed server capacity
• prediction of the limits of the system and how the system will cope as it approaches saturation
• actual figure guidance to other parts of the business on the interest that can be accommodated, e.g. marketing
• identification of areas for concentrated/high quality development

This is typically achieved by using the project specification/design, as well as use of benchmarks of real code runs as different parts of the system are delivered.

What is involved?

All predictions are made using a common methodology that encompasses both mathematical models and simulation. The following stages are followed:

1. define the scope and objectives
2. analyse the system design
3. model or simulate the relevant parts of the system
4. validate the model/simulation
5. execute the model/simulation
6. analyse the result

Where each stage is repeated when necessary to cope with changes in the system and as real benchmark results start to come in. This allows the prediction to be updated when the system components are ready for testing.

Modelling is by use of mathematical formulae based on execution profiles of the functionality investigated. These formulae are typically entered in a spreadsheet to allow simply an easy updating of the model as new data becomes available.

Simulation is performed using Discrete Event Simulation, using an in house, object oriented simulation library. The system is modelled as a set of interacting components represented using java objects.

If you are interested in performance investigation, please contact us for further information.