Tail Sampling is the process of selecting a percentage of the most extremem events or traces in an application, such as errors, timeouts or high latency requests, and discarding the vast majority of other events and traces.
Tail sampling enables you to focus on the outliers, the weird stuff that doesn't happen often, and ignore all the events and traces that are "business as usual". The main goal of tail sampling is to manage the cost of processing and storing all your telemetry data without losing visibility on the anomalies in the datasets. It enables you to zoom in on the rare events that can have a big impact on your system's performance or reliability.
Explore related concepts
Wide event
A wide event is a comprehensive data structure that contains all relevant details of a particular request or transaction within a single event, rather than distributing them across multiple events or logs.
Sampling
Sampling is the process of collecting and analyzing only a subset from a larger dataset, in order to make inferences or observations about the larger dataset. Sampling is about selecting a representative sample that represent the key features of the entire dataset without the cost associated with processing and storing the entire dataset.
Histogram
A histogram a bar chart that shows the distribution of a set of data. It's commonly used in observability to visualize the frequency of values within a dataset.