Features

InteropNet Event Network (eNet) is the network that links vendors, classrooms, demonstration facilities and other systems with the Internet and each other at Interop, the MediaLive International networking shows held in Las Vegas, New York, Paris and Tokyo. The show presents some peculiar problems related to the interoperability of leading-edge equipment, and the short planning and setup times associated with a one-week show. In most other respects, however, eNet is a microcosm of networking in general, presenting the same challenges with regards to security, troubleshooting and management as exist in many corporate LANs.

With 350 cable drops in the main hall and more than 1,000 drops overall, the eNet is comparable in size to many large organizations. In fact, according to Glenn Evans, the lead network engineer for the show, since the network equipment is on loan from 20 different vendors, and there are hundreds of exhibitors connecting their own equipment to the network, eNet might be considered more complex than most corporate networks.

Since each piece of equipment has its own management tools, getting an overall view of network performance becomes more critical and more difficult, as is the ability to remotely poll a variety of devices for information on their configuration, port usage and other statistics.

One system that the network operators used to manage the distributed Interop network is a suite of products from Fluke Networks. This distributed network-analysis solution consisted of OptiView Console, the OptiView Workgroup Analyzer and the OptiView WAN Analyzer.

The Workgroup and WAN Analyzers are collectors that can be placed in each segment of the network, and then report back to the OptiView Console the network conditions at the remote locations. The analyzers communicate with the OptiView Console, which will send alerts if network performance degrades, or when problems occur.

These tools enabled technicians to troubleshoot remote parts of the network without having to leave the network operations center (NOC). At this year’s event, the number of trouble tickets declined from the 300 in 2004 to 180, and average problem resolution time was reduced from 45 minutes to 30 minutes.

There were eight pedestals (i.e., PEDs, or racks of equipment corresponding to a wiring closet) on the show floor, and eight more PEDs at remote locations, each with its own OptiView Workgroup Analyzer. The PEDs held switches, wireless gear, firewalls, voice over IP (VoIP) equipment and more. There were also WAN analyzers deployed outside the Internet router and between the router and firewall.

OptiView Workgroup Analyzers were located between the firewall and primary switch and between the primary switch and the secondary switches. This enabled on-the-spot diagnosis of a variety of problems, all from the single console in the NOC. The eNet managers also used the integrated Web Reporter, which allowed access via browser to console data.

According to Evans, these products can also help determine exactly what is causing problems between various points on the network, providing data on which network devices are generating traffic and where that traffic is going. It also helps identify exactly which piece of equipment is causing a given problem, or if the real problem is between two different devices.

The ability to see a granular analysis of traffic, with trends, bandwidth utilization, even the switch ports in use and what is connected to each port, makes diagnosis of network problems much easier, according to Evans. OptiView Console, in conjunction with the Workgroup and WAN analyzers, can also provide physical configurations of each port, as well as packet captures, and can determine the bandwidth available and network utilization between any two points on the network. It can also generate SNMP traps to alert the NOC in case of problems.

Evans says that being able to get a snapshot of the network at any given time, with historical analysis of alerts, was helpful in the heterogeneous environment. Alerts produced by these tools helped with diagnosis of problems reported by vendors, allowing network operations personnel to confirm that alerts from other systems were correct or not.

While many network-monitoring tools provide lots of data, the challenge for the eNet team was to weed through the mounds of data, looking for changes, trends and problems. The distributed network-analysis products used by MediaLive were able to highlight changes in critical indicators, find problems and identify significant trends, thus easing the task of finding problems without spending hours poring over logs.

In addition to monitoring the network, these distributed products were integrated with management tools from Computer Associates’ Unicenter family of products, passing captured data, accessing trouble tickets and more. eNet techs could remotely access a trouble ticket in Unicenter Service Desk Service Plus using their EtherScope Analyzer, go on-site if necessary, clear the ticket through their handheld device, and get the next ticket without having to return to the NOC. Captured data, including RMON traps from OptiView Analyzers, could also be sent to Unicenter NSM for further analysis and problem resolution.

In order to facilitate collaborative troubleshooting for the network troubleshooting team, each Workgroup Analyzer and WAN Analyzer supported up to eight simultaneous sessions, which meant that multiple technicians could be logged into a remote analyzer at the same time, enabling techs to confer on problems, or to work on separate tasks at the same time.

For more information from Fluke Networks:
www.rsleads.com/511cn-256

Logan Harbaugh is a freelance reviewer and IT consultant located in Redding, Calif., who has worked in IT for more than 20 years. He has also written two books on network troubleshooting.