Artificial Intelligence & Machine Learning

How AIOps Is Transforming IT Operations Management - Part 2

Implementing artificial intelligence for IT operations management, or ITOM, is getting increasingly challenging in light of reduced or stagnant IT budgets and skills shortage. An AIOps solution requires a fit-gap analysis, a strong proof of return on investment and the delivery of tangible business process optimizations, according to a ManageEngine white paper.

See Also: The power of Red Hat Enterprise Linux

In Part 1, we explored the opportunities AI offers for ITOM and other applications. Part 2 outlines the challenges associated with AIOps deployment and how the future of ITOM is shaped.

Gartner projected that market size for AIOps in 2025 will be about $2.1 billion, with a compound annual growth rate of around 19%. While there is optimism and opportunity surrounding AIOps, organizations must address certain challenges to unlock its potential.

"A lot of their clients have big aspirations for building their own AI systems. The problem with that approach is that very few organizations have the skills in-house to build and maintain their own solutions," said Padraig Byrne, vice president analyst at Gartner. "It's a very complex undertaking, requiring very sophisticated skill sets not only in IT operations but also in the areas of data science and artificial intelligence. It also requires resources and capital to derive value for these products."

Skills Shortage

The sheer volume, velocity and complexity of IT operations have surpassed human capabilities. To quickly prevent, identify and resolve high-severity outages and other IT issues, businesses are turning to AI for optimizing IT operations.

"It is difficult to imagine a future for ITOM without AI. The only way to manage complex IT environments to deliver high-performance and scalable services and applications to clients is to use machine learning, AI and AIOps," Byrne said. "It can also reduce the mean time to [problem] identification, thereby shortening the time to resolution. There is no human, scalable solution to the problem."

Integrating AI, machine learning and AIOps requires AI skills and an understanding of machine learning algorithms. This poses a significant challenge for organizations that have already surpassed their IT staffing budgets. The demand for AI talent exceeds the supply, making it a costly resource. Meanwhile, IT environments are becoming increasingly complex. One way to address the problem is to outsource IT management to service providers with the required AI expertise, which also helps reduce costs, fosters scalability and provides access to advanced technologies.

Cultural Setbacks

As with any new technology implementation, gaining user trust during the early stages can be challenging, especially when the system faces "teething problems" as it adapts to an organization's environment. For users, this means learning to use new technology while transitioning from familiar tools or processes.

"IT operations have leveraged [classic] AI, with its mathematical approaches and algorithms, for years," said Carlos Casanova, principal analyst at Forrester. "So the biggest challenges are not necessarily with technology but more with organizational culture. Organizations must accept the latest technologies and tools in the IT operational model."

Marzena Burakowska, principal consultant and partner at Evergo, summarized the security, privacy and reliability aspects associated with implementing AIOps in a LinkedIn blog post. "One of the biggest challenges in adopting AIOps is building trust and reliability in the system. Users must trust the system's decision-making [capabilities] and recommendations to use the solution effectively. Furthermore, most employees base their actions on habits, which is typical for humans," she said.

Security and Privacy

AI relies on large datasets. When AI is integrated into ITOM, there is a risk of sensitive corporate information and personal identifiable information getting leaked. In her blog post, Burakowska said that it is crucial to ensure that the data being used to train AI models is protected, especially as data protection regulations become more stringent. This poses a significant challenge for organizations seeking to leverage AIOps for improved operational efficiency.

"As organizations invest in AIOps, they deal with the need to maintain a delicate balance between utilizing the power of AI in operations and ensuring the security and privacy of their data," she said.

Future of ITOM

Foundational models used for AI will continue to improve and become more versatile, introducing better automation and self-healing capabilities in ITOM. These models will be able to predict failures and use root cause analysis to reduce the mean time to detect them, and they will facilitate faster recovery and remediation.

"What we're going to move toward is a much more hands-off approach to building and scaling large IT services and complex applications, which minimizes - not removes - the direct involvement of a human IT operator inside that environment. That will enable predictive maintenance and self-healing," Byrne said.

AIOps will enable IT organizations to enhance ITOM by providing real-time visibility, detecting patterns based on historical data and acting proactively to ensure secure and resilient operations at the best value possible - with clear cost accounting. "It's a lot to ask for in a solution, but I believe it's closer than most realize. I see aspects of this in action every day. The future of AIOps is happening now," Casanova said.