Skip to content

Commit

Permalink
Introduce "split" metric schema transformation
Browse files Browse the repository at this point in the history
This is a new transformation type that allows to describe a change
where a metric is converted to several other metrics by eliminating
an attribute.

An example of such change that happened recently is this:
#2617

This PR contains:
- A new "split" transformation added to the schema file format.
- Schema file format incremented from 1.0.0 to 1.1.0. This is considered
  a backward compatible change and thus the minor version number is
  incremented.
- Deleted incorrect sentence that claimed that the addition of a new
  transformation type is not backward compatible and requires a
  major version number change. This is incorrect. Addition of
  new transformation types is backward compatible according to our
  current definition of backward compatibility, which is:
  "consumers can consume version X.Y and older versions X.Z
  provided that they are aware of version X.Y".
- Deleted the requirement that a full OTEP is always necessary to
  introduce new transformation types. This seems excessive. I think
  in simple cases like this a PR directly in the spec is sufficient.

This draft PR shows how the corresponding implementation of schema file
parser in Go can be done:
open-telemetry/opentelemetry-go#2999

This commit in my personal repo shows in more detail how the "split"
transformation can be implemented:
tigrannajaryan/telemetry-schema@2b45e48
  • Loading branch information
tigrannajaryan committed Jul 7, 2022
1 parent 2ff1411 commit b71bfa0
Show file tree
Hide file tree
Showing 4 changed files with 58 additions and 17 deletions.
3 changes: 3 additions & 0 deletions CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -56,6 +56,9 @@ release.

### Telemetry Schemas

- Introduce "split" metric schema transformation
([#2653](https://github.com/open-telemetry/opentelemetry-specification/pull/2653)).

### Common

- Introduce Instrumentation Scope Attributes
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -12,8 +12,8 @@ this schema version.
Here is the structure of the Schema File:

```yaml
# Defines the file format. MUST be set to 1.0.0.
file_format: 1.0.0
# Defines the file format. MUST be set to 1.1.0.
file_format: 1.1.0

# The Schema URL that this file is published at. The version number in the URL
# MUST match the highest version number in the "versions" section below.
Expand Down Expand Up @@ -256,6 +256,27 @@ sections. Here is the structure:
# specified below.
```

#### split Transformation

This transformation splits a metric into several metrics and eliminates an attribute.
Here is the structure:

```yaml
metrics:
changes:
- split:
# Name of old metric to split.
apply_to_metric:
# Name of attribute in the old metric to use for splitting. The attribute will be
# eliminated, the new metric will not have it.
by_attribute:
# Names of new metrics to create, one for each possible value of attribute.
attributes_to_metrics:
# map of key/values. The keys are the old attribute value used
# in the previous version, the values are the new metric name
# starting from this version.
```

### logs Section

"logs" section in the schema file defines transformations that are applicable
Expand Down Expand Up @@ -324,8 +345,8 @@ file. The format version follows the MAJOR.MINOR.PATCH format, similar to semver
The "file_format" setting is used by consumers of the file to know if they are
capable of interpreting the content of the file.

The current value for this setting is "1.0.0". Any change to this number MUST
follow the OTEP process and be published in the specification.
The current value for this setting is "1.1.0". Any change to this number MUST
be published in the specification.

The current schema file format allows representing a limited set of
transformations of telemetry data. We anticipate that in the future more types
Expand Down Expand Up @@ -353,19 +374,22 @@ change according to the following rules:
in a backward compatible manner. "Backward compatible" in this context means
that consumers that are aware of the new MINOR number can consume the file of
a particular MINOR version number or of any MINOR version number lower than
that, provided that MAJOR version numbers are the same. Typically, this means
that the added setting in file format is optional and the default value of the
setting matches the behavior of the previous file format version.
that, provided that MAJOR version numbers are the same. This can happen for
example when:

- A new transformation type is added.

- A new setting is added to an existing transformation. The new setting is optional
and the default value of the setting matches the behavior of the previous file
format version.

Note: there is no "forward compatibility" based on MINOR version number.
Consumers which support reading up to a particular MINOR version number SHOULD
NOT attempt to consume files with higher MINOR version numbers.

- MAJOR number SHOULD be increased if the file format is changed in an
incompatible way. For example adding a new transformation type in the
"changes" section is an incompatible change because it cannot be ignored by
existing schema conversion logic, so such a change will require a new MAJOR
number.
incompatible way. This means the consumers of the file need to parse or interpret
the file differently compared to previous MAJOR version.

Correspondingly:

Expand Down Expand Up @@ -426,8 +450,8 @@ To illustrate this with some examples:
## Appendix A. Example Schema File

```yaml
# Defines the file format. MUST be set to 1.0.0.
file_format: 1.0.0
# Defines the file format. MUST be set to 1.1.0.
file_format: 1.1.0

# The Schema URL that this file is published at. The version number in the URL
# MUST match the highest version number in the "versions" section below.
Expand Down Expand Up @@ -527,6 +551,20 @@ versions:
- system.memory.utilization
- system.paging.usage

- split:
# Example from the change done by https://github.com/open-telemetry/opentelemetry-specification/pull/2617
# Name of old metric to split.
apply_to_metric: system.paging.operations
# Name of attribute in the old metric to use for splitting. The attribute will be
# eliminated, the new metric will not have it.
by_attribute: direction
# Names of new metrics to create, one for each possible value of attribute.
attributes_to_metrics:
# If "direction" attribute equals "in" create a new metric called "system.paging.operations.in".
in: system.paging.operations.in
# If "direction" attribute equals "out" create a new metric called "system.paging.operations.out".
out: system.paging.operations.out

logs:
# Definitions that apply to LogRecord data type.
changes:
Expand Down
6 changes: 3 additions & 3 deletions specification/schemas/overview.md
Original file line number Diff line number Diff line change
Expand Up @@ -74,7 +74,7 @@ work together.
Telemetry Schemas are central to how we make this possible. Here is a summary of
how the schemas work:

- OpenTelemetry defines a [file format](file_format_v1.0.0.md) for defining
- OpenTelemetry defines a [file format](file_format_v1.1.0.md) for defining
telemetry schemas.

- Telemetry schemas are versioned. Over time the schema may evolve and telemetry
Expand Down Expand Up @@ -182,7 +182,7 @@ passes through the Collector is converted to that target schema:
## Schema URL

Schema URL is an identifier of a Schema. The URL specifies a location of a
[Schema File](file_format_v1.0.0.md) that can be retrieved (so it is a URL and
[Schema File](file_format_v1.1.0.md) that can be retrieved (so it is a URL and
not just a URI) using HTTP or HTTPS protocol.

Fetching the specified URL may return an HTTP redirect status code. The fetcher
Expand Down Expand Up @@ -215,7 +215,7 @@ Version number follows the MAJOR.MINOR.PATCH format, similar to semver 2.0.

Version numbers use the [ordering rules](https://semver.org/#spec-item-11)
defined by semver 2.0 specification. See how ordering is used in the
[Order of Transformations](file_format_v1.0.0.md#order-of-transformations). Other than the ordering
[Order of Transformations](file_format_v1.1.0.md#order-of-transformations). Other than the ordering
rules the schema version numbers do not carry any other semantic meaning.

OpenTelemetry schema version numbers match OpenTelemetry specification version
Expand Down
2 changes: 1 addition & 1 deletion specification/versioning-and-stability.md
Original file line number Diff line number Diff line change
Expand Up @@ -170,7 +170,7 @@ currently described and are allowed:
- Renaming of span events.

All such changes MUST be described in the OpenTelemetry
[Schema File Format](schemas/file_format_v1.0.0.md) and published in this repository.
[Schema File Format](schemas/file_format_v1.1.0.md) and published in this repository.
For details see [how OpenTelemetry Schemas are published](schemas/overview.md#opentelemetry-schema).

See the [Telemetry Stability](telemetry-stability.md) document for details on how
Expand Down

0 comments on commit b71bfa0

Please sign in to comment.