Proactive network management means heading off problems before they become serious enough to grossly affect end users, such as identifying when a link is becoming so jammed that response time will start to suffer if something isn't done.
So correlation engines, while certainly valuable tools for any organization, don't seem to fit the proactive profile. They root through the myriad alerts that can result from a single device going down and perform analysis that points to the likely source of the problem. By their nature, then, they are reactive, not proactive; the problem has already occurred.
The folks at Tavve Software disagree with that premise, as do some of the company's customers. Tavve's tsc/EventWatch is different because it uses active correlation technology, says Anthony Edwards, the company's founder and chief technology officer.
Say you have a massive outage with 2,000 devices going down, all because a single router failed. But the failed router was just polled and was up at the time. It won't be polled again for maybe five minutes. With most systems, that's five minutes wasted, Edwards says.
Active correlation doesn't work that way. Instead, tsc/EventWatch refers to its network topology database to determine the likely cause of an outage and immediately polls all devices in the suspected path.
"Correlation typically takes one second," Edwards says. With that kind of performance, and depending on the type of outage, network managers have a chance to fix the problem before users notice it. "If you can report a device down before customers realize it's down, that can be considered proactive."
Arun Kant, vice president of IS in the corporate tech-
nology services unit of Prudential Insurance, says Tavve's root cause analysis engine helps his company be proactive.
Prudential has a performance group charged with watching the company's various switches, routers, hubs and other devices to ensure they're operating at optimum levels. If the group sees a router CPU that is overloaded, it can use the Tavve correlation engine to find possible causes.
"Eighty percent of network downtime is spent looking for problems and only 20% is spent fixing them," Kant says. "So a timely root cause analysis is very important. Tavve has helped our performance group work very well and do a proactive type of management."
Keith Ferry, enterprise management technical manager at BB&T, a large bank based in Winston-Salem, N.C., likewise says Tavve helps his company quickly pinpoint the source of a problem when one of his 4,000 network devices goes down.
"You can say it's proactive in that we can find out what's wrong quicker with Tavve than we could before," he says. "Without Tavve, we'd get swamped with a bunch of phone calls and didn't know where the problem was."
RELATED LINKS
From Tavve.
Network fire prevention
You'll need a tool kit that includes performance monitoring gear, application modeling tools and, soon, neural network software. Network World, 9/20/99.
