One practical use-case for using large language models in network operations is to help query large datasets of network telemetry faster and more easily by using natural language. This can improve NetOps and facilitate better data-driven insights for engineers and business leaders. However, the volume and nature of the data types we commonly utilize in networking makes this a daunting workflow to build, hence the dichotomy of build vs. buy.

There are network vendors building natural language query tools for network telemetry, and so it’s important to understand the value, components, and fundamental workflow whether you’re researching vendors or building your own proof of concept.

A reasonable approach, even just to experiment, is to start very simple, such as querying only one data type. That way, we can ensure accuracy and see meaningful results right away. In time, we can build on this initial workflow to include more types of data, perform more advanced analytics, and automate tasks.

Continue reading “Using an LLM to Query Structured Network Telemetry” →