Sidecar or integrated Open Telemetry

Hi,

We have implemented tracing using https://cloud.google.com/trace/docs/setup/python-ot and https://github.com/GoogleCloudPlatform/opentelemetry-operations-python/tree/0b37f8886f8a57b34525631ebd7bb2cf52c51084/docs/examples which is handled inside the container.
By doing it this way it is required to run the Cloud Run instance must be always allocated as stated in https://cloud.google.com/trace/docs/setup/python-ot

So 3 questions:
1) Would implementing open telemetry with a sidecar also require always on instances?
2) What are the benefits of using sidecar open telemetry instead of integrating it right into the api like in https://cloud.google.com/trace/docs/setup/python-ot ?
2) not all our traces are sampled, how can we sample more traces?

Start by reading the linked Google Cloud Trace Python setup and the opentelemetry-operations-python examples. Compare the documented in-container approach with the requested sidecar model and Cloud Run instance allocation, then determine what documentation would answer the sidecar, integration, and sampling questions.

Written by the indexing model from the issue text.