An airline lives or dies on its data network. Aircraft boarding passes, passenger lists and cargo manifests travel across that network, to and from the airline's central data centres to ticket counters and boarding gates in every airport it serves, world-wide.

"If we cannot board a plane or push back from the gate because the network is unstable or unavailable, we have a Severity 1 problem that we need to respond to immediately," says Steve Bingham, network analyst at Phoenix-based US Airways West, formerly America West Airlines. "We work very hard with our ISPs to make sure we do not have those kinds of problems."

Bingham says more than 20 core applications run through the network to support US Airways' highly disbursed, international workforce, which includes sites ranging from Alaska to Peru and west to Honolulu and Maui in the central Pacific. In some of the non-U.S. sites, US Airways West has to depend on third-party providers.

As a result, he says, US Airways West's network includes a lot of redundancy and is based on very fast connections ranging from frame relay up to T1s and T3s. Brute force alone is not enough, however, and Bingham lists the following as major network issues, along with their solutions:

Network visibility: An airline cannot afford to wait for users to start complaining about slow network response times. It needs to anticipate problems and reprovision before they happen. To do that, "we needed a granular view of traffic flows particularly in the application and network layers," says Bingham. "When we reach predetermined capacity thresholds we need to be notified so we can monitor the situation to determine whether we are facing an impending problem or simply need to upgrade that circuit by utilising the forecasting service built within the NetScout appliances."

Managing the website: Airlines are heavily dependent on their websites to provide information and ticketing to both travel agents and directly to a steadily increasing cadre of sophisticated consumers who buy their tickets direct over the Internet. website downtime is costly.

"We first looked at NetScout when we were bringing the America West website in-house from an outside vendor," Bingham says. "Our main concerns were ensuring that the site was up and running 24x7, and to track bandwidth utilisation." And they soon had a problem. "We were seeing excessive bandwidth use by one of our approved vendors, slowing site response time to a crawl." Using the NetScout tools they traced the problem to an application that was performing a large number of screen scrapes.

They were able to contact the vendor and rectify the problem 15 minutes after it was first noticed. "After that, senior management told us, this was the product they wanted to monitor not only our .com and .net, but our data centre and critical sites as well," Bingham says.

Network traffic growth: Today Brigham estimates that the network handles multiple terabytes of traffic a day, and that is growing. "With the constant evolution of new bandwidth dependant applications and steady increase in applications deployed to our field locations, we are looking at alternative network technologies. As of today we use AT&T's MIS. However, we are in the infant stages of designing a new MPLS world-wide network to help carry this and future traffic to our stations. This will allow us to move to quality-of-service/class-of-service-based traffic prioritisation that will prevent HTML and e-mail traffic from overloading our network, for instance, and slowing VoIP and other higher priority packets."

Integrating the networks: Complicating the entire picture is the need to integrate the AmericaWest and US Airways websites and networks. The US Airways East infrastructure is outsourced to EDS, and plans are to bring the integration project in-house, managed from the Phoenix network centre, over the next two years, starting with the websites.

This will double the size of the network and extend it into Europe, making everything more complex. This will create changes mainly in how the network team deals with potential issues, making detailed network monitoring even more important.

"When you start talking about European and other overseas sites we have to work closely with our vendors for redundancy and connectivity," says Bingham. "Today we could send someone from an eastern U.S. location and get on a plane, but we can't do that in Europe. So we have to work more through our vendors."

Security: Like every organisation, US Airways has security concerns, the greatest of which are possible virus and other hacker attacks that could disrupt the network, along with denial-of-service attacks and attempts to hack its website.

"Security is not a huge issue for us," Bingham says. "We rely on Cisco Systems security tools, which do a good job for us." However, they also rely on NetScout to report attacks on the network, which is important for security provisioning.

Overall, says Bingham, "We're confident that we can continue to provide millisecond response to our business-critical applications as we grow. We have to; the success of the business depends on us."