Understanding the Potential of Generative AI in Observability: Exploring Vendor Solutions and Synthetic Monitoring Creation
Introduction
Observability and Generative AI are currently the buzzwords in the IT industry. They’re the subjects of numerous discussions, and an effective Observability strategy should incorporate AI to analyze data, provide insights, automate processes, and simplify the lives of IT professionals. Several Observability vendors are already exploring Generative AI, with some incorporating it into their products.
The primary aim of this article is to shed light on the potential of Generative AI in Observability. We’ll delve into how vendors leverage it and provide a simple analogy for creating synthetic scripts using Generative AI. While we won’t directly explore Observability concepts here (as I’ve already made a comprehensive guide on my Medium blog), we’ll focus on the transformative impact of Generative AI in this field.
What is generative AI?
In the simplest terms, Generative AI is a form of artificial intelligence that can create new content. It leverages large language models that have been trained on extensive datasets of text and code. The transformative potential of Generative AI lies in its ability to revolutionize observability, enhancing its power, efficiency, and user-friendliness.
Some Innovative ideas generative AI can improve observability
Now, it is time to understand how to apply generative AI to some important Observability subjects; I will put here some examples I believe are a perfect match for generative AI.
Anomaly detection: Generative AI models can create models of the expected behavior of systems and applications. These models can then support better anomaly detection, which could be signs of potential problems.
Root cause analysis: Generative AI can generate testing hypotheses about what could be causing the problems and precisely determine the root cause.
Troubleshooting: Generative AI can be used to generate troubleshooting guides documentation or suggest solutions to problems by analyzing the symptoms of the problem and presenting possible causes and solutions.
We can see some of those possibilities applied to real Observability solutions.
How observability vendors are using generative AI
Some of the leading observability vendors are already using generative AI to improve their products. Here are a few examples:
New Relic
New Relic claims to use generative AI in its observability assistant, New Relic Grok. Grok is a powerful tool that enhances observability by:
- Simplifying access to deep insights
- Automating monitoring and optimization processes
- Aiding quantitative decision-making
- Increasing data accessibility
Grok combines large language models (LLMs) and the New Relic unified telemetry data platform to provide deep insights into a system’s state. You can ask Grok questions using a familiar chat interface, and it will respond with in-depth analysis, insights on root causes, and suggested fixes.
Dynatrace
Dynatrace uses Hypermodel AI, a combination of predictive AI, causal AI, and generative AI, to develop a new feature called CoPilot. CoPilot will help users troubleshoot problems more quickly and easily by using generative AI to generate hypotheses about the root cause of issues and suggest solutions.
CoPilot also supports users in creating queries, dashboards, and data notebooks using natural language and generating code for workflow automation. This can simplify the onboarding, configuration, adoption, and usage of the Dynatrace solution.
Datadog
Datadog claims to use generative AI in its Datadog Bits observability copilot, which is designed to help you investigate and respond to incidents more efficiently across the Datadog web app, mobile app, and Slack, without switching contexts.
Datadog Bits enhances observability by:
- Simplifying access to deep insights
- Automating monitoring and optimization processes
- Aiding quantitative decision-making
- Increasing data accessibility
Example on how the generative AI can create automatic Synthetic scripts
One good example of using generative AI in observability solutions is to generate synthetic scripts from real user monitoring (RUM) data. This can simulate user behavior and allow the observability team to customize the scripts based on their specific needs without having to create the monitors from scratch.
This can be useful for testing the performance and functionality of systems and applications before deploying them to production.
Here are some basic steps to achieve this:
- Gather a set of real-world user interactions, such as logins, searches, and purchases, based on the RUM data.
- Train a generative AI model on the collected data.
- Once the model is trained, use a generative AI solution to generate a synthetic script simulating the same behavior collected by RUM.
- Review and edit the script to make sure it meets your needs.
Here are some additional tips:
A good tip is to use a large and diverse dataset of RUM data to train the generative AI model. This will help the model to generate more realistic and comprehensive synthetic scripts, trying to be specific, for example, you can specify the type of user interaction that you want to simulate, as well as the specific steps that the user should take.
Conclusion
Generative AI can revolutionize observability, automating tasks, identifying patterns, making predictions, generating automation and workflows, and creating hypotheses to help observability teams be more efficient and effective.
It is good to see the implementation of Generative AI on some leading observability vendors, like the examples I gave, New Relic Grok, Dynatrace CoPilot, and Datadog Bits, all using generative AI to help users in different ways.
I am excited to see how generative AI will be used to improve observability in the future. I believe that generative AI has the potential to make observability more powerful and accessible to everyone.
Tiago Dias Generoso is a Distinguished IT Architect | Senior SRE | Master Inventor based in Pocos de Caldas, Brazil. The above article is personal and does not necessarily represent the employer’s positions, strategies or opinions.
