Benchmarking/Profiling

[austinlparker]

We could stand to set up a continuous benchmark/profile with the demo. This would involve (to start) a version with no instrumentation and one as stands. We would run these tests after new releases of otel dependencies and publish the results.

[rogercoll]

Just recently, Elastic contributed to OpenTelemetry with its continuous profiling agent. Once the agent is available, I think it could be integrated in the demo to tackle this issue. What are your thoughts on a universal profiler for all the services?

[austinlparker]

yeah, i was thinking we could use the profiler for this.

[austinlparker]

really my motivation here is mostly to provide some sort of starting point/baseline for discussing/thinking about performance and overhead of the sdk and various instrumentations.