Too much traffic in the kitchen.

In working with various routers across the network, the Jack in the Box restaurant chain faced many challenges while monitoring different types of traffic passed from site to site. For example, a WAN network deals with potentially lower speeds than a LAN network. "We're comparing sub-T-1 speeds with Fast Ethernet speeds," explains Jim Antoshak, network support manager. "We needed to quickly view all the WAN traffic that was out there to maximize our pipe."

The IT staff at Jack in the Box oversees a large corporate network connecting headquarters in San Diego with regional field offices nationwide. The company maintains a second network linked via satellite that shares traffic from headquarters to more than 2,000 locations across 30 states. The entire infrastructure contains more than 4,000 nodes and services in excess of 3,400 direct users.

"We separate our corporate traffic from our restaurant traffic," says Antoshak. "Corporate traffic is much more varied and comprehensive and we don't need to make our restaurant traffic over-complicated."

Antoshak decided to invest in a network-monitoring device. After reviewing several options, he selected Network Instruments' Observer Suite package and several remote probes for both company networks. The initial factor in selecting Observer was price, Antoshak says. "I was immediately attracted by the price. Since then, our budgets have increased and I can now afford other solutions, but I'm still satisfied with Observer because of the many features it offers."

Implementing the suite and remote probes proved to be straightforward. Observer was installed on a central console and probes were deployed in various locations. "We had it up and running in about an hour," says Antoshak. "Probes took 15 minutes to set up. Network stats are instantaneous, but for network trending we had to first establish a baseline. A fair sample of data is necessary to create a baseline to measure against. That probably took a few weeks."

The first benefit was the ability to view WAN traffic in total and by specific device. Right away, the team determined there were large amounts of unneeded traffic being sent at inopportune times. "We quickly made our WAN more efficient by controlling what data was being sent and when," says Antoshak. "This was the key aspect to relieving our network congestion."

Many devices can impede network traffic by producing un-needed or unwanted protocols and by producing broadcast storms. Observer can identify and pinpoint these situations. "We were completely unaware that storms were happening on our network," says Antoshak. "Observer immediately identified the storms and showed where the packets were being generated from. With a quick drill-down, we could see that there were too many broadcasts. Our routers were generating too many ICMP redirects and the same packet was being sent twice. Using this information, we logically redesigned our router topology."

Once a baseline was complete, Antoshak compared the network over time to the baseline to recognize unusual activity, growth patterns, congestion and other network issues. Before installing the solution, Jack in the Box did not have a method to determine network traffic levels. Now, using Observer's Real-Time Expert, increased network traffic levels are instantly recognized; and network managers are alerted to network abnormalities and given plausible solutions.

"First I'm notified about issues on particular segments, like broadcast storms or slow response times, through alarm devices that I configure," says Antoshak. "Then Observer provides me with ideas of causes and solutions to the problems. Now, we're much faster at problem resolution. What used to take weeks to resolve is down to a few hours."

The solution also helps the IT team forecast the impact of possible network or application changes without investing in new equipment. The software's "what-if" analysis allows live modeling to predict network bandwidth and response-time impacts for topology changes (e.g., upgrading from 10/100 Mbps to Gigabit Ethernet) or by changes in variables such as average packet size, client-to-server packet ratio, latency, server load and number of users.

"The key benefit with Observer is that instead of throwing bandwidth at our network every time we had congestion problems, we've been able to take our existing capacity and make it more efficient," says Antoshak. "The cost savings have been significant.

"Features like network trending help us to foresee future problems. I don't want users to notify me with issues because I don't want them to see problems to begin with," says Antoshak. "By pushing the existing topology, we're embracing proactive management of our entire network."

For more information from Network Instruments: www.rsleads.com/403cn-256