Sampling is like taking a bite of a delicious cake to understand its taste, without having to eat the whole cake. It enables developers to save data processing and storage cost by gathering and analyzing a smaller portion of data from a larger dataset.
For example, when monitoring user activity on a website, instead of analyzing every single event, we can take a sample of the events to understand patterns and trends.
Instead of collecting and analyzing every single trace from an application, we can select traces that cover all the key features of the application and discarding traces that are very similar one to the other. For example, we could randomly select 10% of all traces but capture all traces where an error occure.
Ultimately, sampling is about reducing the data processing and storage for large datasets at scale.