Part 1 of the Series: AI-Enhanced Infrastructure for AI Ops.

If you ask yourself which are the biggest challenges that your IT model should address, endless answers will surely come to your mind. But the majority will find a common theme: your organization faces a manually intensive and complex environment that demands access to information and resources with even greater agility and flexibility.

The emergence of hybrid work, multicloud, edge computing, containers, IoT, 5G, and ever more significant security threats are all contributing to this scale and complexity multiplier. All this at a time when the user experience that the network delivers has never been more critical to most organizations adopting digital processes at their very core.

Through multiple conversations I have had on this subject with customers and experts in the field, I identify the application of artificial intelligence (AI) for IT operation (AIOps) as one of the most relevant transformations within the IT industry. With this in mind, I’m writing this blog series based on my own experience and discussions with experts and customers. My objective is to lay out the role of AI in the infrastructure domain and how AI-enhanced infrastructure is a critical foundation for a broader AIOps strategy. Along the way, I want to dispel the myth that AI is the panacea for everything. It is hugely powerful for some use cases – and less so for others.

So, let’s get started at the top – with AIOps.

What is AIOps?

The term AIOps, initially coined by Gartner, stands for “artificial intelligence for IT operations.” Broadly speaking, we can understand AIOps as the use of artificial intelligence capabilities, such as big data, machine learning (ML), and machine reasoning (MR), to solve IT challenges within and across traditional operational silos such as NetOps, DevOps, SecOps and Cloud Operations.

Ultimately, the AIOps role is to deliver the best possible user experience by assuring consistently reliable, high-performing, and secure services with maximum efficiency and speed. This means applying AI to activities like availability and performance monitoring, operations planning and provisioning, network assurance, event correlation and analysis, troubleshooting and remediation, and the automation and management of IT services.

An AIOps platform makes it possible to make sense of the ever-growing amount of data generated within IT operations, analyzing the information to identify relevant performance patterns and presenting it in a constructive way, aimed towards more simple and agile reactive and, in some cases, predictive decision making.

I would like to share six aspects that I consider relevant when evaluating your own IT infrastructure transformation path to drive an AIOps model:

1. Now is the right moment for AIOps

In the past several years, ITOps and NetOps teams have increased the adoption of AI/ML-driven capabilities. In many cases, the path to fully leverage these investments, as happens with many emerging technologies, is not entirely clear. Capitalizing on the value of AI requires an investment in new knowledge and skills for optimal implementation.

But why now – we might ask. One reason is because we can. The cloud and the Internet have provided us with the infrastructure – high-power compute and network resources – to aggregate and process enormous amounts of data economically.

But the more compelling reason is that the scale and complexity of managing IT infrastructure without artificial intelligence to help us make good and speedy decisions are no longer realistic. Consider that according to IDC, by 2025, there will be 55.7 B connected devices worldwide, 75% of which will be connected to an IoT platform1. By 2026, 90% of Global 2000 CIOs will use AIOps solutions to drive automated remediation and workload placement decisions, including cost and performance metrics, improving resiliency and agility.2

Familiarizing yourself with AI technologies, capabilities, and language is critical for any IT leader. Even if a project focused on AIOps is not in the immediate future, you may find a need for it sooner than you anticipate.

2. AIOps demands an integrated approach to your Infrastructure strategy

An AIOps model requires the integration of the various existing IT platforms, tools, and processes, unifying access to information, insights, and capabilities that were previously managed in separate silos.

An organization will not get the full benefit of AIOps if its technological systems cannot share information and learn from each other. Especially considering the vast amounts of structured and unstructured data found in cloud data lakes that demand continuous and resilient integration approaches to provide real-time insights.

As part of an adoption strategy, the organization must consider integration methods of data across domains, which is possible thanks to data virtualization and federation technologies and the usage of application programming interface (API) and software development kits (SDK).

3. AI-enhanced infrastructure can be a competitive advantage

With AIOps, activities such as network remediation could be solved in just minutes instead of hours. While it is tempting to evaluate the benefits to IT only in terms of cost reduction, there is a considerable amount of advantage in how these technologies can improve organizational agility and efficiency while also reducing operational risk.

The financial benefits should also be evaluated on how AI enables new business processes that would either be too resource prohibitive or prohibitively complex. For example, this could include identifying emerging customer trends and needs, securely deploying IoT applications and devices at scale or enabling new edge computing enabled services.

4. AIOps means focusing IT talent on higher-value activities

IT Ops and Net Ops teams spend considerable time on control, monitoring, and management tasks that could be automated, which explains why over 50% of network strategists identify AI as a priority network investment.2

One of the most significant barriers to adopting automation in ITOps is the belief that systems will replace humans. It is important to note that it is unthinkable for IT staff to manage the scale demanded by today’s IT operations without automation. But also consider how ITOps automation allows IT and networking teams shift effort and talent to focus on developing the new sets of skills and roles required as part of the new AIOps model.

5. The evolution to a “trusted” closed-loop AI-enhanced model

Due to the accelerated adoption of cloud-first models in today’s IT environments, it is evident that networks need to deliver consistent performance between highly distributed users, devices, applications, and workloads. The opportunity is to streamline the complete process with AI-enabled visibility and insights informing automated remediation actions to keep the network continuously aligned to service level requirements. For example, let’s say an ML-enabled predictive analytics capability identifies a potential impending Internet performance issue and informs the controller of a recommended remediation via an alternate service provider. For the foreseeable future, administrators will not trust the automatic remediation, and manual approval/activation steps will be required. However, once enough positive experience and trust have been gained, it is conceivable that the closed-loop process will be fully automated.

Network Analytics

6. Be practical on where to start

Let’s remember rule number one. AI is a means — not an end. It is unlikely that your organization is keen on experimental scenarios to “dip their toes” in AI adoption. The most likely path to success is to start small by identifying the areas that are eating up a lot of resources or holding you back from supporting the business. These are potential opportunities for introducing AI capabilities that can help your IT team while also providing an opportunity to identify and close any skill gaps. While you build experience with AI use cases, you can also develop a strategic AIOps roadmap that allows you to eventually bridge across operational silos towards a streamlined AI-enabled IT operations model.

The path to an AI-enhanced infrastructure

AIOps is no longer a utopian nor a far-fetched paradigm. More and more organizations are adopting IT strategies that facilitate the transition to AI-enhanced ITOps and NetOps. Still, its roadmap requires evaluating real-world use cases that defy traditional infrastructure approaches while generating a larger sense of urgency for adoption. Businesses from all industries can start reaping the benefits of optimized IT and network operations while enabling innovation through intelligent data consumption across their entire organizations.

This is something that I intend to explore in subsequent blogs. In the meantime, I would love to hear about your own approach and perspectives – How do you think an AI-enhanced infrastructure could help with your IT/Network operations?


[1] IDC Press Release: IoT Growth Demands Rethink of Long-Term Storage Strategies – 2020
[2] IDC FutureScape: Worldwide Future of Digital Infrastructure – 2022 Predictions
[3] Cisco 2020 Global Networking Trends Report


Pradeep Kathail

Chief Network Architect

Intent Based Networking Group