Newfangled applications such as streaming media, Web conferencing and voice over IP promise lots of benefits, but they also can suck up network bandwidth to the point where these and other programs are practically paralyzed.
Application performance is a big concern and it's getting bigger, said more than 180 Network World readers who responded to a not-for-profit e-mail survey we conducted in conjunction with bandwidth management vendor Packeteer (complete survey results can be obtained here). Roughly two-thirds said performance of their applications has degraded significantly, and about 43% said slowdowns increased during the past 12 months.
The effect of application-performance degradation varies, but network IT executives said loss of productivity among individual employees and teams is a big concern. They also said poor performance can interfere with revenue-producing activities and hurt customer service.
What's more, one in three respondents said their company has held off on deployments of ERP, CRM, portal and other applications because they were concerned about performance.
Identifying the problem
One challenge in addressing performance issues is that problems involve sanctioned and unsanctioned applications, the latter of which include Web surfing, peer-to-peer file sharing and instant messaging.
Another issue is that companies often don't have a good handle on just which applications are on their networks. Only one in four respondents said they know precisely what applications are running and how much bandwidth they consume. Nearly 39% said they estimate that unsanctioned programs account for between 21% and 40% of total traffic.
When it comes to which part of the IT infrastructure is at the root of application performance problems, there's plenty of blame to go around. Asked to rate the importance of four areas in this regard on a scale of 1 to 5 (with 1 being not important and 5 being very important), here's what respondents said:
• Servers: 4.07.
• Applications, architecture deployment: 4.07.
• LAN network: 3.80.
• WAN network: 3.75.
Finding a fix
Despite growing concern about application performance, most companies appear to be stuck in a reactive mode, addressing problems only after end users and customers pipe up. Even with network monitoring tools in place, a staggering 72.6% first learn about performance problems from end-user calls to the help desk, and another 82.3% said employee complaints usually are the first they hear of slowdowns on their networks.
"Most of the systems we've had typically were telling us what happened after the fact. We wanted new systems to give us a bit of warning," says Mordecai Sabo, a systems engineer with the distributed infrastructure unit in the technological bureau at the Israeli National Police in Jerusalem.
The organization recently began rolling out new features in Computer Associates' Unicenter network and systems management software that Sabo says will help the network team improve the reliability of critical applications. "The software can help us understand better what happens when a server goes down, and to know what else is affected by the outage," Sabo says.
Not surprisingly, close to two-thirds of respondents said they are somewhat, less than and/or not at all satisfied with their current network monitoring tools.
Network executives said they are taking a variety of actions to ensure applications run smoothly. Adding bandwidth ranked as the most popular fix, with about 60% of respondents saying they planned to beef up their WAN capacity to handle more-demanding applications.