Status: Stable
Table of Contents
This section applies to Resource, Span, Log, and Metric attribute names (also known as the "attribute keys"). For brevity within this section when we use the term "name" without an adjective it is implied to mean "attribute name".
Every name MUST be a valid Unicode sequence.
Note: we merely require that the names are represented as Unicode sequences. This specification does not define how exactly the Unicode sequences are encoded. The encoding can vary from one programming language to another and from one wire format to another. Use the idiomatic way to represent Unicode in the particular programming language or wire format.
Names SHOULD follow these rules:
-
Names SHOULD be lowercase.
-
Use namespacing to avoid name clashes. Delimit the namespaces using a dot character. For example
service.version
denotes the service version whereservice
is the namespace andversion
is an attribute in that namespace. -
Namespaces can be nested. For example
telemetry.sdk
is a namespace inside top-leveltelemetry
namespace andtelemetry.sdk.name
is an attribute insidetelemetry.sdk
namespace. Note: the fact that an entity is located within another entity is typically not a sufficient reason for forming nested namespaces. The purpose of a namespace is to avoid name clashes, not to indicate entity hierarchies. This purpose should primarily drive the decision about forming nested namespaces. -
For each multi-word dot-delimited component of the attribute name separate the words by underscores (i.e. use snake_case). For example
http.response.status_code
denotes the status code in the http namespace. -
Names SHOULD NOT coincide with namespaces. For example if
service.instance.id
is an attribute name then it is no longer valid to have an attribute namedservice.instance
becauseservice.instance
is already a namespace. Because of this rule be careful when choosing names: every existing name prohibits existence of an equally named namespace in the future, and vice versa: any existing namespace prohibits existence of an equally named attribute key in the future.
-
When an attribute represents a single entity, the attribute name SHOULD be singular. Examples:
host.name
,container.id
. -
When attribute can represent multiple entities, the attribute name SHOULD be pluralized and the value type SHOULD be an array. E.g.
process.command_args
might include multiple values: the executable name and command arguments. -
When an attribute represents a measurement, Metric Name Pluralization Guidelines SHOULD be followed for the attribute name.
Abbreviations MAY be used when they are widely recognized and commonly used.
Examples include common technical abbreviations such as IP
, DB
, CPU
,
HTTP
, URL
, or product names like AWS
, GCP
, K8s
.
Abbreviations SHOULD be avoided if they are ambiguous, for example, when they apply to multiple products or concepts.
A new attribute MUST NOT be added with the same name as an attribute that existed in the past but was renamed (with a corresponding schema file).
When introducing a new attribute name check all existing schema files to make sure the name does not appear as a key of any "rename_attributes" section (keys denote old attribute names in rename operations).
-
All names that are part of OpenTelemetry semantic conventions SHOULD be part of a namespace.
-
When coming up with a new semantic convention make sure to check existing namespaces (Semantic Conventions) to see if the new name fits.
-
When a new namespace is necessary consider whether it should be a top-level namespace (e.g.
service
) or a nested namespace (e.g.service.instance
). -
Semantic conventions exist for four areas: for Resource, Span, Log, and Metric attribute names. In addition, for spans we have two more areas: Event and Link attribute names. Identical namespaces or names in all these areas MUST have identical meanings. For example the
http.request.method
span attribute name denotes exactly the same concept as thehttp.request.method
metric attribute, has the same data type and the same set of possible values (in both cases it records the value of the HTTP protocol's request method as a string). -
Semantic conventions MUST limit names to printable Basic Latin characters (more precisely to U+0021 .. U+007E subset of Unicode code points). It is recommended to further limit names to the following Unicode code points: Latin alphabet, Numeric, Underscore, Dot (as namespace delimiter).
Note: Semantic Conventions tooling limits names to lowercase Latin alphabet, Numeric, Underscore, Dot (as namespace delimiter). Names must start with a letter, end with an alphanumeric character, and must not contain two or more consecutive delimiters (Underscore or Dot).
As an application developer when you need to record an attribute first consult existing semantic conventions. If an appropriate name does not exists you will need to come up with a new name. To do that consider a few options:
-
The name is specific to your company and may be possibly used outside the company as well. To avoid clashes with names introduced by other companies (in a distributed system that uses applications from multiple vendors) it is recommended to prefix the new name by your company's reverse domain name, e.g.
com.acme.shopname
. -
The name is specific to your application that will be used internally only. If you already have an internal company process that helps you to ensure no name clashes happen then feel free to follow it. Otherwise it is recommended to prefix the attribute name by your application name, provided that the application name is reasonably unique within your organization (e.g.
myuniquemapapp.longitude
is likely fine). Make sure the application name does not clash with an existing semantic convention namespace. -
It is not recommended to use existing OpenTelemetry semantic convention namespace as a prefix for a new company- or application-specific attribute name. Doing so may result in a name clash in the future, if OpenTelemetry decides to use that same name for a different purpose or if some other third party instrumentation decides to use that exact same attribute name and you combine that instrumentation with your own.
-
The name may be generally applicable to applications in the industry. In that case consider submitting a proposal to this specification to add a new name to the semantic conventions, and if necessary also to add a new namespace.
It is recommended to limit names to printable Basic Latin characters (more precisely to U+0021 .. U+007E subset of Unicode code points).
Attribute names that start with otel.
are reserved to be defined by
OpenTelemetry specification. These are typically used to express OpenTelemetry
concepts in formats that don't have a corresponding concept.
For example, the otel.scope.name
attribute is used to record the
instrumentation scope name, which is an OpenTelemetry concept that is natively
represented in OTLP, but does not have an equivalent in other telemetry formats
and protocols.
Any additions to the otel.*
namespace MUST be approved as part of
OpenTelemetry specification.