Sidebar: Predicting Disaster

Computerworld - For Mike Kennedy, the value of a capacity planning tool lies in its ability to predict the effect of a discrete event on future needs. Kennedy, vice president of the infrastructure performance management group at J.P. Morgan Chase & Co.'s data center in Hicksville, N.Y., faced such an event when his company acquired a credit card portfolio from San Francisco-based Providian Financial Corp. in 2002.
Kennedy is in charge of anticipating capacity on the company's Cisco-based network, its call center system and some 1,000 midrange Unix/Windows/AS400 servers. He uses Opnet's IT Guru suite, which pulls data from CiscoWorks, HP OpenView and other management tools. The IT Guru software stores performance and configuration data in a virtual network environment so simulations can be run without re-creating the data each time.
Kennedy put IT Guru to work modeling what would happen when an additional 3.5 million Providian credit card customers were suddenly dumped into J.P. Morgan's call center and production environment. Some effects were obvious; others were not. For example, the trunks leading to the voice-recognition units were inadequate to handle the increased traffic, and J.P. Morgan was able to avert a disaster before the Providian customers came online.
"If we didn't do that modeling, we would have had an outage. It would have cost us $1.2 million per day in losses," he says.
Kennedy now models all changes before they go into production. "Eliminating problems before they happen has a tremendous influence on the bottom line," he says.

Mike Kennedy of J.P. Morgan Chase & Co.