Case Study: T-Mobile Network Performance Monitor 2                                

Project Type: Networked Application
Customer: T-Mobile

Product Overview:

Following the success of an initial application from Penrillian, T-Mobile returned with a brief to extend the software. Demand for the new application was high based on the success of the initial trial, so the brief came with a very tight deadline.

"Penrillian has delivered a highly effective application that has a measurable impact on the service we provide to our customers."

The first application sampled a series of web pages and then delivered the results to T-Mobile via email. This meant that an individual had to collate reports and circulate them to interested parties within the company - a time-consuming task.

When T-Mobile wanted to extend the application to more than 40 handsets, the task would have become impossible.

Objectives:

 - To continue to test the 'real-life' service customers will experience
 - To automate the testing process as much as possible, making it very user friendly
 - To collect results in an efficient, easy-to-analyse fashion

Method Deployed:

Penrillian redeveloped the original application to provide for remote management and centralised data-storage. Each handset can now be remotely configured to tweak the test cycle, varying the web pages it looks at, and other test parameters. For example, the handsets might be instructed to download a music track, or attempt some other mobile internet function. Instead of emailing the results, each handset now logs in to a centralised database and uploads its most recent results.

Penrillian proposed an incremental deployment to ensure that the project could deliver what was required within the tight deadline.

Results:

Penrillian completed the development on schedule and also managed the rollout, providing first-line support for any problems. The application has now been rolled out to 45 handsets, permanently deployed in T-Mobile stores around the UK. Running continuously since Christmas the system has been running tests on multiple URLs, including the BBC, Google and eBay homepages, every 15 minutes of every day, collecting more than 2 million results.

All of the information has been stored on a central database where it can easily be analysed by T-Mobile's engineers and cross-referenced with results from their own in-network tools. At present the system is being used to test the impact on loading time of various tweaks to the Web'n'Walk homepage - just one of a variety of uses for the data.

"The new system provides an extremely cost-effective way of measuring real-world performance of T-Mobile's network. Each handset added to the system costs just one percent of the price of a typical in-network node," commented Monty Weekes, T-Mobile. "The results enable us to make sound decisions about new network investment. We have already upgraded coverage in two sites based on the feedback from the system, ensuring that our customers receive the best possible service."

Click here to download this case study in PDF format