Cloud computing, virtualization and mobility increasingly dominate the conversation among IT professionals.
But an industry analyst says the role of the network is not to be underestimated.
The network plays important role in optimization of applications and how they perform, Zeus Kerravala, senior vice-president of research at Yankee Group, told a Toronto audience on Thursday.
“The cost of computing continues to drop, the role of the network continues to grow,” he said, “and I think the network does become the most cost-effective, scalable platform for apps and services.”
He was speaking as part of a 10-city North American seminar sponsored by network equipment maker Cisco Systems Inc. and Netscout Systems Inc., which makes network performance management tools. Officials from both vendors also spoke at the half-day session about how their products fit into network trends.
As networks become platforms for delivering applications as a service, Kerravala said, they have to continually be optimized and monitored. That means the way the network is managed has to change as well.
“It’s not thinking about fault management and what device is up or down, because we’ve built networks so redundant today you can lose a device and it really doesn’t affect it. What you have to understand is what the normal operating parameters are, and then when you reach a certain deviation that should alert you there’s some kind of problem.”
On the other hand, he said, legacy tools aren’t sufficient for real-time network management. Static maps are common on such tools, he said, but the environment isn’t static any more. Probes and agents are good for troubleshooting but not for ongoing monitoring, he added.
The largest cause of network downtime – 37 per cent — is human error, Kerravala said. [“When I was a network engineer I probably contributed to that number,” he confessed.] They survey also showed that 90 per cent of the time administrators spend resolving a problem is identifying it. If you want to find a way of doing more with less money, he concluded, find ways of finding problems faster.
Kerravala also had advice for managing major applications on the network.
Voice over IP (VoIP) and video are unique applications, he argued, that many organizations have little experience with. VoIP, for example, isn’t bandwidth intensive but is sensitive to jitter, latency and packet loss. That can make troubleshooting difficult.
Use of video will undoubtedly increase with more devices – like tables and smart phones – having cameras.
“You want to think about UC (unified communications) having its own life cycle where you are continually watching the network, understanding how traffic’s flowing,” he advised, “and then continually optimizing it.”
Thinking of the network as a control point for security can create a more scalable security environment, Kerravala also said. Understanding how your network normally performs means any suspicious activity gets immediately noticed.
Desktop and server virtualization brings their own problems, but with real-time network analysis tools will show sudden shifts in network traffic patterns to identify problems, he said.
“We are on the verge of another inflection point in IT,” he concluded, “driven by consumerization, multimedia, virtualization and mobility. You want to think about shifting some of your spend away from legacy management tools and look at things that can give you more of a service delivery look at things.”
Automate operations as much as possible, he advised, particularly mundane processes that can cause human errors. Focus on tools that can help drive down problem identification time, he said, which makes network managers more pro-active than waiting for users to complain about problems you don’t know about.