IT operations are faced with rapidly growing volumes of data generated by infrastructures and applications. Distributed in "disconnected silos", these teams find it difficult to manage the most urgent incidents. To prevent, identify and resolve them more quickly, companies are turning to artificial intelligence (AI). AIOps is a concept created by Gartner in 2016. Four years on, few companies have adopted it. But the situation is changing. A fusion of AI and Ops (as in operations), IAOps can therefore be defined as the use of artificial intelligence to manage IT operations.
AIOps
Hence the interest shown by IT managers. Gartner forecasts that the exclusive use by large enterprises of IAOps. to control applications and infrastructures will rise from 5% in 2018 to 30% in 2023.
Simply put, IAOps is the application of machine learning (or Machine Learning-ML) and data science to IT operations problems.
IAOps platforms combine big data and ML functionality. To enhance and partially replace all primary IT operations functions, including availability and performance monitoring, event correlation and analysis, and IT service management and automation.
The aim is therefore to provide global visibility (by collecting and correlating data from disparate sources) on IS performance, and to automate as many IT operation flows as possible.
No company can afford IT problems that last long enough to impact on the user experience. But the human ability to react is insufficient, and automation becomes a necessity.
IT technicians and engineers can no longer keep up with the pace of technology stack maintenance. Manual management is no longer tenable, and it takes one machine to manage another.
Beware of preconceived ideas: IAOps does not eliminate the need for human intervention in the maintenance and enhancement of corporate information systems. Rather, it reduces it.
Indeed, IAOps introduces capabilities to automate routine preventive and corrective activities, and to display data visualizations that IT technicians can interpret and use for more comprehensive and faster interventions if required.
AIOps: advice from Gartner
For many companies, the time has come to integrate IAOps. In order to launch a sustainable IAOps initiative, Gartner offers a number of tips:
- Don't wait: familiarize yourself with the vocabulary and capabilities of AI and ML today, even if an IAOps project isn't imminent. "Priorities and capabilities are changing, so you may need them sooner rather than later," warns the firm;
- Choose your first test cases wisely: transformation initiatives benefit from starting small, capturing knowledge and repeating from there. Adopt the same approach to successfully integrate IAOps;
- Develop and demonstrate your skills: it's essential to demystify this concept, which is still too abstruse, to your colleagues and managers. Simple techniques should be presented, gaps in skills and experience identified, and a plan drawn up to fill them;
- Experiment freely: IAOps platforms are still too expensive and complex a route to familiarize yourself with these solutions. Instead, Gartner recommends starting with the many open source ML software ;
- Look beyond IT: leverage the data and analytics resources that may already be present in your organization. Data management is an important component of IAOps, and teams are often already skilled ;
- Standardize where possible, modernize where practical: it's essential to prepare your infrastructure to support an eventual IAOps implementation by adopting a consistent automation architecture, code-based infrastructure and immutable infrastructure models.
IAOps in conclusion
In conclusion, IAOps bridges three different IT disciplines. Service management, performance management and automation to achieve its objectives of continuous knowledge and improvement.
In accelerated and "hyperscalar" IT environments, there needs to be a new approach that takes advantage of advances in big data and machine learning to overcome the limitations of tools and people.
