Using Advanced Network Monitoring and Analytics to Tame Holiday Traffic Spikes
Not so long ago, network costs spiked alongside holiday and other peak period demands. That’s because IT primarily sought to add capacity to accommodate the extra network traffic. Close network monitoring of traffic was necessary to respond to spikes, and thus all the buzz was about scaling to traffic and elasticity in the network. But today that approach, when used alone, can be costly.
“On average, companies spend 70 percent of their IT budgets just to maintain the current network and ‘keep the lights on,'” NetworkWorld stated, reporting on IDC data. “The ability to reduce hardware maintenance and data center support without compromising uptime is appealing, especially as organizations address IT megatrends encompassing security, mobility and the Internet of Things.”
Hiccups in Network Monitoring and Traffic Shaping
However, maintaining the network daily and periodically scaling it to fit dips and peaks in its use isn’t enough to ensure high-priority traffic gets through unimpeded.
Traffic shaping, or packet shaping, is a method of delineating priority levels for types of data moving over the network and has been traditionally used to direct priority traffic to the front of the line. It has been helpful in ensuring that important information transfer and work is unimpeded, primarily by pushing less important traffic to the end of the queue.
However, as historically used, this tactic, too, has its limitations, not the least of which is a limited number of classifications. This can be inefficient since it lacks granular controls.
Machine Learning Moves to the Head of the Class
A research paper published by the IEEE describes how new classification methods developed through machine learning have improved traffic shaping in recent years. Those efforts are ongoing. Even so, traffic shaping can only do so much to help ease network traffic jams because the traffic itself is growing at a phenomenal rate.
Simply put, ever-growing traffic loads pile into bigger jams, no matter how orderly they are stacked. Hence the IEEE’s conclusion that “traffic classification performance can be improved significantly” — and not that traffic classification fixes everything!
Obviously, more must be done to keep the network at peak performance and eliminate lag and downtime during periods of high demand. The first order of the day then is to shoot for true and deeper visibility in the network in order to better manage packet flow and system performance. Such action requires advanced networking monitoring.
Get Proactive With Analytics
Equally crucial to meeting peak demand is the use of proactive fault management, which is to say the use of predictive analytics with machine learning to identify and predict faults and performance degradation within the network. This enables you to make the network far more efficient, which in turn limits the need to expand in order to scale until you actually need to do so.
The beauty of this new class of analytics is that machine learning detects, or “learns,” the normal operational behavior of your network. In doing so, it is able to detect problems and anomalies long before humans think to look for them.
It is also capable of analyzing network performance and data monitoring across silos, domains and vendors. In effect, it becomes an invaluable watchdog as it is always on, always watching and always noticing the slightest irregular tick as soon as it happens.
Further, machine learning improves over time since each day’s network events teach the system even more.
Best Network Play
In short, the adoption of new technologies is driving enterprise networks to the breaking point. Adding large loads from holiday and other peak periods is only a further strain.
Continuing to use old methods in network monitoring and maintenance to try to keep up is a losing proposition. The best course of action is to move to advance network monitoring with proactive analytics as quickly as possible.