Around 50 per cent of IT and network issues are still reported by end users rather than being discovered by the IT department first, says Motti Tal, executive vice-president of marketing, product and business development for OpTier Inc., a New York maker of management software.
The goal continues to be to reduce that number.
The more critical IT and networks become to business the more desirable it becomes to see the problems early – ideally before they affect end users. But the more complex networks get, the harder it is to do that. There are too many interdependencies. Even when something is clearly wrong, determining the root cause can be daunting. And knowing where to look for signs of problems that aren’t yet obvious is even trickier.
“The challenges are centred around having a highly redundant system with lots of components,” says an IT manager at a large financial company who asked not to be identified. “It becomes challenging to trace activity through a redundant, parallel infrastructure.”
“I give it the tag line ‘death by metrics,’” quips Doug Roberts, business development manager for Everett, Wash.-based Fluke Networks’ performance management business. “You have more data than you could possibly consume.”
The broad outlines of how to solve this problem are largely agreed upon. Most vendors and network managers talk about the need for a complete, integrated picture of the IT environment and a top-down view of what’s going on. According to research firm International Data Corp., integration to provide network management through a “single pane of glass” continues to be a dominant theme. You need to start with what end users see – the applications, services and transactions – and then quickly get down through the layers to the source of the problem.
“What (customers) want is a single light or a metric for a service that says the service is up and running,” says Scott Sobers, global program director for the communications sector at IBM Corp.’s Tivoli management software business. “We are looking for something which is telling us what’s the reason and how to fix it,” says Tomasz Kunicki, chief executive at New York-based AdRem Software, Inc., maker of NetCrunch monitoring software. Monitoring tools need to do more than alert network managers to the symptoms, he says. They need to indicate what’s wrong and what needs to be done to fix it.
Dashboards don’t do it
Network management vendors started trying to address this with management dashboards several years ago, but Tal says dashboards don’t really do the job. The problem, he says, is that dashboards require IT experts to map the structure of the network and how it relates to applications before they can give a meaningful overview of what’s going on.
“Dashboards assume that you have an expert tell you how different components are related to the applications,” Tal says. OpTier starts with a view of what end users see – the actual transactions going through the network. From this, OpTier’s Business Transaction Management software traces problems down to specific hardware or software components. Beyond helping resolve problems, Tal says, business transaction management can also be a tool for understanding how applications are used. “It doesn’t have to be an outage tool,” the financial firm IT manager says.
“We use it to kind of extract out performance challenges that we’re finding over time and then iron them out of our infrastructure.”
In their own ways, most network management tools vendors are trying to address the need to see the big picture and then drill down to technical details. It requires not only presenting information so it is easy to understand, but pulling together data from multiple vendors’ tools. That is coming, gradually. “We spend a lot of time building and developing partnerships,” says IBM’s Sobers.
Pressure on IT departments to reduce the number of tools they use is also pushing smaller vendors to ally with the large system management companies, IDC says. Tools vendors are also emphasizing the need to address problems before they become visible. Corey Copping, product marketing manager for ProCurve Networking products at Hewlett-Packard (Canada) Inc. in Mississauga, Ont., gives wireless access points as an example. Increasing wireless use, and the emergence of voice over IP phones that run on wireless networks, have increased the risk of congestion on wireless access points, so HP’s ProCurve Manager Plus software includes the ability to monitor traffic on individual access points and move connections to adjoining access points to balance the loads and help fend off congestion problems.
VoIP adds complexity
Voice over IP not only adds traffic to the network but have different management needs than traditional traffic, IDC notes. Because voice is so sensitive to delays and network problems are instantly and painfully visible to people using VOIP phones, ensuring availability and low latency have become ever more critical. Here too, says Roberts, administrators need tools that can assess the quality of an individual phone call and then drill down to identify the causes of any problems.
Fluke recently added VOIP-specific features to its MetroScope Carrier Ethernet Analyzer. Because VOIP handsets often draw their electricity from the network, they add the need for Power Over Ethernet technology, which is another aspect of the network to be managed, Copping observes.
Some organizations are moving beyond VoIP to video, either for desktop videoconferencing or by moving surveillance video for building security from its own dedicated network to sharing the same network as data and voice. This further increases both traffic and the need for Power Over Ethernet.
Security is still number one
And while more security cameras are being connected to data networks, mention security to most network administrators and they will think first not of cameras but of securing the network itself.
“Security continues to be the number-one issue on top of every IT manager’s mind,” Copping says. Network-management tools have a significant role to play in watching for security breaches as well as technical glitches.
Server virtualization adds another twist. Managing applications in a virtualized environment first requires keeping track of where they’re running, notes Jacques Nadeau, solutions director for infrastructure management at Islandia, N.Y.-based CA, Inc. The fact that an application can move easily from one server to another complicates the problem of drilling down from an application issue to the root cause.
So as tools evolve in an effort to meet network managers’ needs, the evolution of technology continually introduces new challenges. “There’s never going to be network management, network monitoring utopia,” Roberts admits.