Mark "Instrumentation Units" and "Instrumentation Types" sections of the general metric semantic conventions as stable

[trask]

Fixes #3291
Closes #705

Changes

Marks Instrumentation Units and Instrumentation Types sections of the general metric semantic conventions as stable.

[github-actions]

This PR was marked stale due to lack of activity. It will be closed in 7 days.

[github-actions]

This PR was marked stale due to lack of activity. It will be closed in 7 days.

[bogdandrutu]

Please rebase

[trask]

I'd suggest merging (or rejecting) #3312 first before merging this

[jack-berg]

@reyang brings up a good question about our unit recommendation for utilization metrics, which we say should use 1 as the unit. This language seems inspired by UCUM but not necessarily consistent with other recommendations we make about unit. For example, we say that unit should be specified using case sensitive ("c/s") variant of unit, but 1 doesn't meet this criteria.

[trask]

@reyang brings up a good question about our unit recommendation for utilization metrics, which we say should use 1 as the unit. This language seems inspired by UCUM but not necessarily consistent with other recommendations we make about unit. For example, we say that unit should be specified using case sensitive ("c/s") variant of unit, but 1 doesn't meet this criteria.

I think 1 is an important unit for a lot of cases, e.g. it's the implicit unit for {request}.

[trask]

@reyang brings up a good question about our unit recommendation for utilization metrics, which we say should use 1 as the unit. This language seems inspired by UCUM but not necessarily consistent with other recommendations we make about unit. For example, we say that unit should be specified using case sensitive ("c/s") variant of unit, but 1 doesn't meet this criteria.

I think 1 is an important unit for a lot of cases, e.g. it's the implicit unit for {request}.

oh, I see your point, {request} is a c/s, e.g. {rbc} at https://ucum.org/ucum#section-Examples-for-some-Non-Units.

but maybe not 1 itself(?)

[trask]

Once #3312 is merged, I will add a note to this PR covering @jack-berg's #3312 (comment):

We could add a bullet to this PR which makes recommendations by domains. Something like:

* When measuring durations, seconds (i.e. `s`) SHOULD be used.

[jack-berg]

@trask I opened #3388 to address using seconds for durations. This allows us to keep the scope of this PR isolated to changing the stability status.

[trask]

@reyang brings up a good question about our unit recommendation for utilization metrics, which we say should use 1 as the unit. This language seems inspired by UCUM but not necessarily consistent with other recommendations we make about unit. For example, we say that unit should be specified using case sensitive ("c/s") variant of unit, but 1 doesn't meet this criteria.

I sent #3393 to try to resolve this