There are two reasons behind this proposal:

• To create less ambiguity when naming new metrics
• I think it creates slightly better attribute names (when sharing the metric namespace), e.g.
  • system.process.status (vs system.processes.status)
  • jvm.thread.state (vs jvm.threads.state)

This would affect these metrics names:

• system.processes.count -> system.process.count
• system.processes.created -> system.process.created
• db.client.connections.* -> db.client.connection.*
• jvm.threads.count -> jvm.thread.count
• jvm.classes.count -> jvm.class.count

It may also affect these metrics if we decide to add .count (or .usage) to them:

• process.threads -> process.thread.count
• process.open_file_descriptors -> process.open_file_descriptor.count