Glossary · Observability
What Is a Span?
Written by
Amber Jain
Published
May 2026
Reading time
4 min read
A span is one timed operation in a trace, and how spans nest tells you the structure of a request.
A span is the basic unit of work in a distributed trace: a single named operation with a start and end time, a set of attributes, a status, and a link to its parent span. Spans nest to form a complete trace.
What a span contains
A span records a name (the operation), a start and end timestamp (its duration), attributes (key-value context such as the HTTP route or database statement), a status (ok or error), and references to its parent and trace. Events and links can add further detail.
Root and child spans
The first span in a request is the root span; every operation it triggers becomes a child span, and children can have their own children. This parent-child nesting is what turns a flat list of operations into a tree that shows how a request fanned out and where the time went.
How it fits Opstral
Sentinel AI reads span attributes and status when it investigates, using the error span and its parent chain to localise a fault to a specific operation.
Frequently asked questions
What is a root span?
The root span is the first span in a trace, representing the entry point of the request; all other spans in that trace descend from it.
What are span attributes?
Key-value pairs that add context to a span, such as http.request.method or db.system, following OpenTelemetry semantic conventions so they are consistent across tools.