Laurent - stock.adobe.com
Automation gets things done more quickly, but AI-infused automation gets things done before humans can figure out what's needed.
AI's rapid growth has brought tremendous value to organizations across a wide range of verticals, including the IT sector. Software vendors use machine learning, a subcomponent of AI, to solve what are traditionally difficult issues, including those related to automation and anomaly detection.
AI for IT operations, also called AIOps, is in its early days. In the coming year, you will see more references to terms like AI automation, automation via machine learning and smart automation. Understand what they mean for IT security, cloud resource provisioning and more.
Automation can replace human output with fewer errors and more efficient task completion. It frees us from the mundane and repetitive steps to achieve a goal, the things that keep us from more creative, intuitive projects. But it has limits.
Automation requires a predefined set of triggers and actions. For example, a script can provision a VM faster than a person can, or a robot can attach a car door in a factory with exactly the same precise movements repeatedly. In these and other examples, automation completes the work thanks to the if-then-else computer language construct: If a VM with the name of SRV1 doesn't exist on X VMware cluster, then create it. If the car door is not attached to the frame, then ascertain what model it is and then connect that door.
This kind of automation requires humans to know all of the rules upfront. It asks us to predict the future and lay out all of the tasks for computers to perform when given certain events. Most IT automation available today falls into this dumb category, but what if it could be smart? Expect the next wave of automation in the form of AI for IT operations in 2019.
Examples of AI for IT operations
While automation has dramatically helped IT organizations deliver better software, it still relies on us to build all of the triggers and rules ahead of time. AI for IT operations means we automate the automation, not the tasks. Two examples illustrate the value of AIOps, in terms of anomaly recognition for secure operations and algorithm-based decision-making for cloud scaling.
Thousands of events happen every second on enterprise networks. Most of the traffic is legitimate, but nefarious acts can and do happen. Security software vendors' anomaly detection capabilities in antivirus software still depend on virus definitions, which are rules set by humans. Vendors must continually update and maintain these rules for the product to stay effective. Smart automation with AI driving decisions could intelligently handle each event and distinguish what's legitimate from what's nefarious.
Security is a huge area for AI for IT operations, given the risk involved with security breaches and the fast-changing threat landscape. However, AI also can benefit organizations in practical areas, such as infrastructure deployment sizing.
Consider the IT infrastructure supporting the customer-facing website of an online retailer, which is under heavier-than-normal use throughout the holiday shopping season. Today, online retailers set thresholds, such as CPU utilization or storage space, with their cloud hosting provider and then adjust operations based on those thresholds. In 2019, AI-based capacity planning applications could be trained to detect the precise time when customer demand increases, necessitating these resource adjustments, and instruct the cloud infrastructure service to change settings. This intelligence-directed automation could save organizations money on cloud costs and deliver optimal service levels to their customers.
The options are limitless for AI in IT operations, which I call smart automation, in 2019. If vendors can convey the concept of smart automation through AI to organizations, we will see explosive growth in this field.