When it comes to AI and managing your network, this old Johnny Mercer song describes what using AIOps (AI for IT Operations) boils down to today. In fact, the first few lines sum it up very accurately. It seems that everything old is new again…
“You've got to accentuate the positive and eliminate the negative, latch on to the affirmative, don't mess with Mister In Between…”
Why Eliminate the Negative (Reactive Tasks)?
Reactive tasks are very disruptive. They impact the business, interrupt normal activities, and delay planned activities. Minimizing reactive tasks and their impacts reduces the time it takes to resolve problems which equates to maximizing ITs efforts. AIOps improves the user experience and minimizes the impact issues have on the business.
A survey we conducted in 2021 identified that 50% of an IT person's time is spent on reactive tasks. Half of someone’s day is spent problem solving and answering questions about when the problem will get fixed. No time spent on learning something new or moving that new project forward.
Reactive tasks take on two forms that IT must deal with: all out failures and poor user experience issues. Although failures can contribute to a poor user experience, I like to look at them differently.
- A failure is when something just does not work and the problem it creates tends to be easier to find and remedy. Some examples:
- An endpoint client cannot connect
- Services are not reachable - AAA, DHCP, DNS, etc.
- An application does not work at all
- A networking component stops working - AP, switch, gateway, link, tunnel, etc.
- A poor user experience is when things work, but not as intended. These problems are more difficult to resolve because they tend to be intermittent and hard to find. Some examples:
- Web pages take 15 seconds to load
- Collaboration apps dropping and reconnecting
- In a call while roaming and losing a few seconds of audio
- Video call where the image freezes
From the user perspective, both are annoying. From an IT perspective, they can require very different troubleshooting steps and skill sets.
The good news is that AIOps helps with both types of problems by minimizing the time it takes to identify, triage, troubleshoot, and resolve them. AIOps Automation helps reduce the time it takes to resolve a problem from days or hours to minutes in a lot of cases. We have customers mentioning 50-to-90%-time savings for some tasks. And some of their newer or less experienced team members are contributing at a much faster rate.
Why hunt for days to fix problems related to possible failures caused by the following when AIOps can narrow your focus:
- AP and user issues that are really happening due to switch port errors and flaps
- The same goes for PoE issues in remote locations that users are dealing with but have never opened helpdesk tickets for
In fact, a good AIOps solution that includes dynamic baselines and peer comparisons should recommend configuration or infrastructure changes based on best practices before a problem is reported.
The Proactive Nature of a Good AIOps Solution
While AIOps is used to reduce the time, it takes to resolve issues, it is better to eliminate pesky help desk calls if possible. The goal is to provide a frictionless experience for both the IT admins and end users by predicting a failure before something happens. Here are some examples of where AIOps is well suited:
- Detecting suboptimal firmware – security vulnerabilities, fixes, new features
- A fan in a switch is underperforming – there’s a temperature spike, an increase in power consumption, and eventually failure
- APs are not connected to a switch stack for optimal coverage if a switch goes offline
- Avoiding an RF coverage hole that users will experience
AIOps is not only useful for finding or predicting problems, but is also excellent at helping make a good experience great through providing optimization guidance based on measured best practices.
- Transmit power optimization
- Transient client pruning
- WAN uplink optimization
Even though we have been using AIOps for years to solve and optimize customer’s networks, we are still at the beginning of our journey. And while we may not be able to detect every problem before it happens, we are going to try. We are listening to our customers, reviewing TAC cases, and adjusting our AI/ML models to not only fix common problems, but things you will hopefully never have to experience – like the completely new phone or IoT (Internet of Things) device that takes down APs or will not connect reliably.
Learn more about AIOps and the Aruba AI portfolio.
Related resources
