The third leg of the stool, Leary says, is automation of both the analysis itself and the actions that derive from it. Automation of network analytics differs from predictive analytics in that it is not predicting when problems might occur, but merely looking at current performance.
“Using data and analytics to see how a network is performing has been around for quite some time, but we’ve noticed that the amount of data being generated today is getting harder to process, as there is just so much more,” says Trent Fierro, senior marketing manager for cloud and AIOps at Aruba, a Hewlett Packard Enterprise company. “The ultimate goal is to help IT organizations spot anomalies and trends that provide network, security and application insights that also help the business offer new services to its users while improving existing services.”
What Is Predictive Network Analytics?
If network analytics focuses on using data to look at past performance (and, hopefully, to monitor the current state of the network in real time), predictive network analytics is squarely concerned with anticipating what the future state of the network might be.
Using historical data fed into machine learning models, predictive network analytics looks at network components, resources and the end-user experience to extrapolate when a business might need to do an upgrade or “when you may need to reconfigure the network to avoid a problem,” Leary says. The key is to avoid the problem before an IT team needs to mitigate it, he says.
With network analytics, network managers can see if there is a problem on a network link and change a configuration to achieve better performance, Compton says. However, he adds, that is “fundamentally a reactive response,” in contrast to predictive analytics, which looks at patterns of network behavior to predict future operating scenarios.
For example, there might be a pattern of network overload on Monday mornings as workers all log in at the same time to download email and fire up cloud-based applications. Predictive analytics can anticipate this and give operations teams the ability to take proactive steps to avoid the issue by, say, routing traffic to different network links in advance of the expected spike in traffic.
While not a perfect crystal ball, predictive network analytics lets IT teams “make that change proactively” to deliver a better user experience, Compton says.
Why Are Network Analytics and Predictive Analytics Important?
Ordinary network analytics data is useful in providing IT teams greater visibility into the state of their network and where problems are cropping up. However, the approach is predicated on a reactive mindset.
As organizations move toward a more predictive model, their IT teams gain efficiencies, experts say. Not only can resources be positioned ahead of time, but by helping IT teams avoid network issues instead of reacting to them, predictive analytics eases the load on IT help desks.
“It really is a force multiplier,” Compton says. “If they could do something as a planned cadence, as opposed to an emergency ticket coming in, it is a radically different level of productivity.”
Instead of focusing on the mean time to repair and shortening the duration of an incident, he says, “all of a sudden you’re in a mode where you’re actually avoiding the incident in the first place.”
“The use of AI and machine learning open the door to a variety of new use cases that stem from finding endpoint connectivity issues to how applications are performing on a network, even as traffic traverses the cloud,” Fierro says. “Traditional analytics usually focuses on a specific task, whereas predictive AIOps [AI for IT Operations] solutions work across your entire network — wireless, switching, WAN, security and applications — to provide a consolidated view.”