'Ping' is the key to managing 3M's growing network - use of echo testing in management of local area networks - Data/Voice Test Equipment

Communications News, May, 1992

For Mark Anderson, a network

< design specialist at

manufacturing giant 3M, the biggest challenge in developing an integrated, enterprise-wide network from 80 isolated LANs (local area networks) scattered throughout the United States is in providing proactive management. "Before, if one LAN went

< down, we could wait until it was up again because it affected only those few users," says Anderson, who manages 3M's extensive X.25 network with over 1,000 nodes. "Now with the whole company interconnected and using centralized information resources, a local network's down-time can effect our entire operation." Anderson provides proactive management

< by testing remotely for node activity and periodically measuring network responsiveness. He uses Hewlett-Packard's LanProbe. The key to its expanded management ability lies in the new functionality of the ping test, or echo test. "Today's echo test goes beyond letting

< me know if a node is dead or alive," Anderson explains. "Now echo test alerts me to potentially hazardous situations so I can deal with a problem before it effects users."

                   <
 Prior to the X.25 environment, 3M's
                   <
network consisted of a LAN for each
manufacturing, customer service and
sales site, along with a corporate network.
Each of the sites managed their
own LAN, fixing problems as they occurred.

                   <

If they couldn't fix it, they would call Anderson. Anderson is currently transforming the < independent LANs into an integrated client-server environment. "Now if a critical node goes down, 50

< users tap their fingers until the node is running again," Anderson notes. Clearly, 3M's rapid growth in automation < and connectivity dictates a need for network management to prevent downtime. Installing 10 HP-4990S LanProbes on

< critical segments provided Anderson with the right tools for proactively managing critical nodes and segments from his own office. Remote monitoring was a key consideration for configuring a client-server environment spread across the country. If a controller calls saying he cannot

< retrieve manufacturing data, Anderson can determine instantly where the problem is by performing a quick echo test on the client PC and server nodes. Even if it were cost effective to send a technical specialist to the PC site, if the problem is actually with a server located 400 miles away in Aberdeen, South Dakota, the trip would be a waste. Besides one-time echo testing, Anderson

< regularly monitors critical nodes. "A background echo test goes out every

< hour to insure that all servers and any client nodes which need to access the server are operational," he explains. Since nodes that change in status turn

< from green to red, Anderson immediately attends to the problem before users are affected. Statistical data gathered by background

< tests is tabulated into a spreadsheet format, giving Anderson exact information on when and how long a node is up or down. "This helps identify trouble

< spots," Anderson explains, "so I can make intelligent decisions for re-configuring or expanding the network based on solid facts." In addition to providing an echo

< test for determining whether a node is dead or alive, the Lan-Probe echo test records the echo's return time and monitors which protocols are running on which nodes. An initial broadcast echo test

< alerts Anderson to which devices are running which protocols. After that, background testing will alert him to protocol changes. "If I set up a LAN running only TCP/IP, < and in the next background test I see other protocols running, I know there has been a change. LanProbe shows exactly where the unauthorized node is so I can alert the local network administrator." The unit's echo test "response time"

< feature gathers statistics, generating minimum, maximum and average ping return times through the background testing and exports data for analysis in spreadsheet format. Anderson can determine base-line response times for echo tests and monitor changes over time. "When dramatic changes occur in response

< time," Anderson says, "I know immediately to look into what might be causing the problem. Perhaps there is too much traffic for that LAN's bandwidth, so I might re-route a segment or increase the available bandwidth. In any event, if left unattended, the problem will compound. "With the multi-featured ping test, I

< have more ways to troubleshoot and make sure the network runs correctly," Anderson says. Now his only dilemma is that 3M's

< network is running so well that no one really believes that managing the network is a challenge.

                   <