AIOps is great if you need to remediate relatively simple and straightforward problems that arise in the context of monitoring and management.
Since the dawn of modern software monitoring, reacting in real time to application problems has been the holy grail of IT teams everywhere. Until recently, however, it was a feat that few could achieve. There was always a lag between when monitoring systems exposed a problem, and when engineers could take action. Thanks to AIOps, however, that is finally changing.
The technology is poised to enable true real-time application monitoring and management, bringing a once-unimaginable level of efficiency to this facet of IT operations.
See also: AIOps Gaining Traction as Technology Accelerator
The application monitoring dilemma
The application monitoring and performance management strategies that IT teams use today haven’t changed fundamentally since the late 1990s. That’s when teams started deploying tools like Nagios, which monitor software environments for anomalies or outliers, then send alerts to engineers so that they can investigate and rectify the issue.
The problem with this strategy is that it’s not possible to react in real time. No matter how efficient your monitoring tools and alerting workflows are, there will always be a delay in response when relying on human engineers to solve problems.
Enter AIOps
AIOps is on the path to change this. The acronym stands for Artificial Intelligence Operations or (depending on whom you ask) Artificial Intelligence for IT Operations and refers to the use of AI and machine learning to solve IT operations problems or assist in IT operations workflows. Gartner coined the term in 2016, but it was only in the past year or two that vendors like Broadcom, Splunk, and Instana have embraced the concept and started unveiling software monitoring tools that incorporate AIOps features.
The core idea behind the technology is that, by using AI to make decisions instead of relying on humans, IT teams can take instantaneous action. If a server is running out of disk space, an AIOps tool could allocate more storage automatically and in real time, for example. If an attacker is launching port scans against a network, AIOps could block the offending host instantly.
In these ways, AIOps is poised to bring true real-time operations to application monitoring and performance management. AIOps offers a range of benefits for other IT workflows, too, from infrastructure provisioning to data lifecycle management. But it is particularly valuable for IT teams who have traditionally suffered from an inability to solve application availability and performance problems in true real time.
Limitations to consider
That said, it’s important to recognize that AIOps can’t make every type of software monitoring or management task happen in real time. There will always be situations where humans need to become involved in the process, and those situations entail delays.
AIOps is great if you need to remediate relatively simple and straightforward problems that arise in the context of monitoring and management. Updating infrastructure configurations, firewall rules, access control-policies, and so on is what AIOps excels at because those workflows can be scripted ahead of time.
When you get into more complex issues, however, AIOps isn’t a ticket to real-time remediation. A cyberattack that originates from a range of addresses and leverages a variety of exploits is not something that AIOps can shut down with the flick of an automatic switch, for instance. Nor can AIOps solve problems that involve physical hardware failure or necessitate the rebuilding of a software environment from the ground up.
These limitations are especially important to bear in mind, given that AIOps is fast becoming one of the leading buzzwords of 2021. Expect to hear more about it and its enormous potential going forward. As you do, recognize that AIOps can bring real-time functionality to application monitoring and performance management, but only in some cases. AIOps is a powerful solution, but it shouldn’t be over-hyped.
