IT has dreamed about autonomous IT networks for a decade. While a single all-in-one solution may not exist yet, the pairing of two innovative technologies — SASE and AIOps — points toward potential breakthrough innovation in network management.
When AI is applied to the problems of IT operations, it’s referred to as AIOps. Machine learning, behavioral analytics and predictive analytics are used to observe and evaluate performance across the network, security and cloud applications. Using data science, AIOps visibility engines recommend ways to optimize performance, and the engine has the potential to act nearly instantly on its own recommendations. When it does, this is known as closed-loop automation, marking the key difference between partial automation and a fully automated AIOps system or autonomous network.
With the ability to automate manual processes that have plagued companies for decades, AIOps is expected to completely revolutionize IT the same way that desktop computing and the Internet did in years past. When humans can’t crunch network data fast enough nor pinpoint the root cause of an application outage in today’s multi-cloud environments, AIOps steps in with an answer and a prescription to put the right bandwidth in the right places at the right time. It can even predict when network and security performance will not meet expectations. That means that organizations that automate more elements of their network management processes are able to deliver services faster—and suffer fewer outages.
For instance, AIOps may tell you that video conferencing bandwidth needs are growing and that consumption will saturate your network within the next three months. Moreover, it can show you where to boost bandwidth, the optimal network path certain applications should take, and how to adjust application policies and settings to ensure a high quality of service. With closed-loop automation, network management changes would be made for you.
But the full potential of AIOps’ can only be realized when its engine has what it needs.
AIOps engines are powerful tools, but without the right environment, IT investments fail miserably. The challenges with AIOps are threefold:
This is where secure access service edge (SASE) solutions can make the perfect pairing for AIOps because it solves all three problems with one “stone.” This category of solutions combines software-defined wide area networks (SD-WAN) and security capabilities in one service. When these converged solutions act as the underlying technology platform for AIOps, the AI engine has everything it needs.
In order for AIOps to have the transformational impact IT leaders are looking for, it needs to observe and engage with a wealth of data streams, including information about the network, security, cloud services, applications and users. It needs:
However, pairing SASE and AIOps doesn’t solve every problem. The engine needs time to learn the behavior of its new environment and fail at its own attempted problem-solving. IT managers need to take time to coach it. Smart networks may know to alert to hardware failures, but they will still need to be taught finer skills. For example, it may not understand what a surge in website traffic means during December and how to adjust accordingly. Training periods are key for making the AI engine more effective at its job as well as for building trust between humans and machines. After all, automation isn’t possible if AIOps isn’t coached and trusted to act alone.
As SASE unlocks the power of AIOps, a new breakthrough in autonomous networking is unfolding. The self-managing, self-healing networks that industry leaders have forecasted for years are revealing themselves. It’s time for forward-leaning IT executives to start paying attention because autonomous networks should be hitting the market in the first half of this decade.Comcast Business can help enterprises navigate their digital transformation journeys with global secure networking. Learn more today.
See how AIOps and SASE work together to enhance netowrk management and visibility