It's almost like magic, but predictive network technology is anything but a trick.
Using artificial intelligence (AI) and machine language (ML) mathematical models and algorithms, predictive network technology alerts an organization to network issues as early as possible and offers problem-solving solutions. "The technology enables networks to learn from past instances using massive amounts of data through predictive analytics," explains Titus M, a senior analyst with technology and business research firm Everest Group. "It collects network telemetry data, recognizes trends, and forecasts network difficulties that might negatively impact user experience and offers potential solutions to the issue."
Predictive network technology can also suggest network remediation solutions for automatic or manual implementation, depending on the use case, at the discretion of the IT networking or operation team, says Sam Halabi, technology consulting competency leader at business advisory firm EY.
Predictive network technology's value is that it helps network operations transition from a reactive to a proactive model when it comes to addressing potential issues. "Network problems can happen due to many factors, such as degradation in the transport network, bandwidth congestion/traffic loss, suboptimal routing, network outages, and so on," Halabi says. "Such problems are very disruptive to the business and can have a major negative financial impact when they occur."
Challenges and opportunities
Although a powerful and beneficial tool, predictive network technology presents some serious risks. One concern is that the system can only make decisions based on available options. "If you haven't planned for it or have not trained it for certain situations, the system might not be able to respond appropriately," says Chuck Everette, director of cybersecurity advocacy at Deep Instinct, a cybersecurity technology company. Everette reports he has witnessed situations "where automated decisions were happening at such a pace you couldn't make heads or tails of the root cause due to the constant changes in the adaptation of the network trying to fix or heal itself."
Potential adopters should also be aware that predictive network technology only functions effectively when adopted by both enterprises and their service providers. "Predictive network technologies must work end-to-end to predict and remediate problems," Halabi says. "Enterprises that adopt the technology might be hindered by having a service provider that doesn’t use the technology." He advises potential adopters to check with their service providers for possible compatibility issues.
New adopters also face the challenge of deciding whether to hand their network operation over to AI software. When predictive network technology is in the driver's seat, human managers and operators will still be held accountable for any wrong decisions the software makes on their behalf.
Halabi advises taking a middle-of-the-road approach to adoption. "As with any new tool, operations personnel will likely try to make the best of the [service] recommendations given by [predictive network] systems but will probably not do an immediate transfer to automatic remediation," he says. "The path to success will be a combination of efficient predictive technology software combined with skilled IT and operations teams who have extensive experience."
A good way to get started with predictive network technology is to select a solid use case–a pain point or other critical business need–and then run a trial to see how things work out. "Plan it like you would any Agile project," suggests Michael Haynes, principal client engagement leader, global telecommunications industry at IBM. "This is a sensible approach because the insight gained often needs to be validated or triaged by a person before it is automated," he explains. While technology can scale, people can't. "By starting small, teams are more likely to show incremental progress, learn as they go, and guide the next logical direction of the plan," Haynes says.
Large enterprises and telcos are already positioning themselves as early predictive network technology adopters. "Medium, small, and micro-enterprises will follow suit once the technology is proven to operate reliably," Haynes predicts.
As networks evolve and grow increasingly complex, the automatic remediation offered by predictive network technology promises to reduce management burden while leading to better performance and fewer service outages. While many network technologies, such as SD-WAN, have already moved toward optimizing traffic routing based on network and application performance and visibility, such technologies remain reactive to a certain extent. "Predictive network technologies are game changers because they address both problem avoidance as well as problem remediation," Halabi says. "This technology is well positioned to become a baseline for the emerging software-defined and application-aware network."
The key thing to remember about predictive network technology is that it doesn't eliminate the need for human monitoring and oversight. "This type of technology is not 'set-and-forget-it,'" Everette states. "It needs constant monitoring and tuning in order to stay at peak efficiency.”