AI in DevOps For More Efficient Infrastructure Management and Monitoring

Which types of AI do DevOps teams use? 

Artificial Intelligence refers to computer systems programmed to carry out tasks like learning and problem-solving. By feeding on swathes of data, AI can solve challenges, reduce waste, and discover new efficiencies. 

 

In the field of DevOps, AI quickly becomes an indispensable part of task automation, reducing error and providing teams with insights for more informed decisions.

 

The most common type of AI in DevOps is Large Language Models (LLMs), which presents an immense potential for tasks like writing code, testing, and deployment.

 

Let’s now focus on how engineers take advantage of it in more detail. 

AI in DevOps: six use cases and relevant tools

#1: CI/CD pipelines

Continuous integration and development processes are DevOps’ bread-and-butter, so this list couldn’t start with anything else. 

 

One particularly valuable use case for AI in CI/CD is testing your code for security. ML models can quickly understand the logic of your code and find potential vulnerabilities. One of the pioneers in this field is Snyk with its DeepCode AI, which you can integrate into your app building and testing.

 

AI is also getting better at checking your code for quality. Devs can write reliable code faster with ‘AI coding assistants’ often available as plugins in popular IDE environments like GitLab Duo or Github Copilot. 

 

Artificial Intelligence is also helpful for deploying code to production. For example, Harness uses ML to analyse logs from systems to see how your app’s performance changes after deploying a new build. 

 

Of course, there are many more ways to reap AI’s benefits for CI/CD, so I’ll discuss them in a separate blog post soon. 

 

#2: Anomaly detection

Anomaly detection is one of the most popular AI applications in DevOps. 

 

This capability is possible thanks to monitoring and analysing data from sources like logs, metrics, and events to spot deviations from usual patterns. DevOps teams can detect potential anomalies by setting alerts before they snowball into full-blown issues.

 

AI-powered anomaly detection is now part of popular monitoring platforms like DataDog and Splunk; all key cloud providers also offer similar functionalities.  

 

For example, Amazon has an entire portfolio of anomaly detection tools for software and hardware infrastructure. Amazon DevOps Guru and Lookout for Metrics are great for detecting operational and business data inconsistencies. 

 

Regarding detecting and remediating vulnerabilities, Amazon offers CodeGuru together with CodeGuru Security.

 

To detect anomalies in the hardware and physical infrastructure, consider Monitron and Lookout for Equipment. The first one analyses your equipment’s vibration and temperature, while the latter automatically analyses industrial equipment’s sensor data to detect abnormal behaviour and avoid downtime. 

 

#3: Issue diagnosis and analysis

AI-powered solutions support DevOps teams in diagnosing and analysing issues. Once Machine Learning algorithms detect anomalies, they can identify patterns and correlate them with events across entire IT systems, no matter their complexity. 

 

Tools like Dynatrace AIOps, New Relic, and Datadog Trace Queries use AI mechanisms to diagnose issues in real time, provide insights into root causes, and suggest remediation strategies.  

 

These platforms analyse volumes of telemetry data to identify performance bottlenecks, config errors, and security vulnerabilities to help DevOps teams resolve system issues swiftly and reliably.

 

#4: Infrastructure optimisation

Another DevOps area in which AI shines is infrastructure optimisation. This capability is possible thanks to analysing usage patterns, predicting demands, and automating resource allocation. 

 

AI-powered tools can optimise your cloud infrastructure by dynamically adjusting resources to match workload fluctuations, ensure efficient utilisation, spot waste, and allow proactive adjustments.

 

This type of support is available, for example, on Google Cloud Platform under the umbrella of Active Assist. GCP’s tools help you optimise cloud operations by recommending ways to reduce expenses, boost performance, improve security, and even sustainability. 

 

By analysing your usage, Active Assist suggests the most optimal type of VMs, ways to rightsize them, and flags underutilised resources. 

 

#5: Context-aware predictions and recommendations

Microsoft’s primary offering is Copilot for Azure, which can support you in everything from writing code to infrastructure consultations in the Azure dashboard. 

 

This product direction shouldn’t be surprising considering Microsoft’s partnership with OpenAI, the company behind ChatGPT and the subsequent GenAI craze. 

 

However, Amazon’s Q proves that the Redmond-based giant does not hold a monopoly on Large Language Models and AI-powered assistants.

 

#6: Network security and intelligence

Network security and cloud environment have been vital in the cloud computing era due to increased network data transmission and the global distribution of resources.

 

AI supports DevOps in network security by enabling faster anomaly detection, predictive analysis, threat intelligence, and automated response. These capabilities are possible thanks to analysing network traffic patterns and predicting and identifying threats from vast data sources. 

 

One of the popular tools in this segment is Prisma SD-WAN. By integrating AI-driven security features into network management, the solution helps to strengthen network defences and mitigate security risks.

 

Another tool is VMware Edge Network Intelligence, which uses AI to analyse and automate edge network jobs. Predicting issues and automating management and security tasks boosts your DevSecOps efficiency and reliability. 

 

After months in preview, IBM Hybrid Cloud Mesh became available at the end of 2023, providing support in network traffic optimisation.

 

Another interesting case in this category is the detection of unauthorised access. This paper showcases how to use hybrid deep learning models to detect and prevent acts of intrusion on AWS. 

 

Considering the hype surrounding the R&D and business of AI in all industries, we can expect even more innovative applications to come shortly. AI offers almost endless opportunities, so the arms race is on. 

Why AI in DevOps won’t replace engineers (at least, not yet!)

Despite an accelerating pace of AIOps development, it doesn’t yet pose a threat to human engineers. 

 

While AI handles tedious and repetitive tasks gracefully, greatly accelerating software development, DevOps embraces so much more than just automation. Human engineers remain pivotal in system design, implementation, and using AI-generated insights for optimisation and strategic decisions. 

 

Even the most innovative tool will never replace collaboration, problem-solving, adaptability, and other soft skills that are the foundations of DevOps culture. With the right AI solutions in place, your engineers can focus on higher-value tasks and use their expertise to the fullest extent.

 

So instead of fearing Gen AI, it’s time to embrace it as a complement to your DevOps team’s work—and now is the perfect time to act.