You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
#128 introduced service topology of instrumented nodes. One issue with this is that an application that makes calls to remote services such as database and LLMs show up as a monolith and nothing else:
I'm not sure how other visualization products work, but possibly we can handle "db.system" or "gen_ai.system" to allow leaf spans to indicate who they are calling. I think there's a "peer.service" attribute, but it seems not in use in normal auto-instrumentation (at least in python)
While in the future this, ".system" thing may change, it gives a stable more usable bucket than hostname which is problematic, and more visually interesting.
#128 introduced service topology of instrumented nodes. One issue with this is that an application that makes calls to remote services such as database and LLMs show up as a monolith and nothing else:
I'm not sure how other visualization products work, but possibly we can handle "db.system" or "gen_ai.system" to allow leaf spans to indicate who they are calling. I think there's a "peer.service" attribute, but it seems not in use in normal auto-instrumentation (at least in python)
While in the future this, ".system" thing may change, it gives a stable more usable bucket than hostname which is problematic, and more visually interesting.
Notes:
https://opentelemetry.io/docs/zero-code/java/agent/instrumentation/#peer-service-name
Consistent naming:
db.system
todb.system.name
, namespace constants, removedb
from system-specific names open-telemetry/semantic-conventions#1734The text was updated successfully, but these errors were encountered: