OpenTelemetry or Spandex for DataDog inegration

Hey folks, I want to migrate an app from Appsignal (still love ya) to DataDog, for the purpose of better tracing and integration with logs.

Should I be looking at BEAM’s implementation of OpenTelemetry or that’s not quite there yet? I don’t actually care too much about other providers at the moment, we’re probably going to stay with DataDog for a good while, so the platform-independent aspect of OpenTelemetry isn’t a big concern.

I am leaning towards just using Spandex, or should I give OpenTelemetry a try instead?

1 Like

Maintainer of spandex said that OpenTelemetry is a “long-term replacement for tools like Spandex”.

Here is an example of an application that uses OpenTelemetry to report traces to DataDog.

So, I’d suggest to give OpenTelemetry a try.

5 Likes

I definitely think OpenTelemetry is the better option as Datadog, Honeycomb and perhaps AppSignal support it ? Maintainer of Spandex mentioned OpenTelemetry will be a long-term replaced, especially since it’s supported by EEF Observability workgroup :wink:

2 Likes

I myself is currently trying to use opentelemetry for purpose of tracing with jaeger, my current impression is:

  1. long-term opentelemetry is the standard & future
  2. that being said, currently opentelemetry is pretty new (just recently get to 1.0) so there’s not a lot of documentation/example and sometime you would need to get greasy reading code & debugging strange issue (for example, i recently found issue regarding missing doc for grpc config if you are using grpc on opentelemetry_exporter additional doc for config grpcbox for grpc protocol by calvin-kargo · Pull Request #366 · open-telemetry/opentelemetry-erlang · GitHub )
  3. that being said, for jaeger at least, it’s seems even though opentelemetry is new, there’s already a lot of support for in in ecosystem (phoenix, ecto) compared to alternative like ex_ray or fretchen
1 Like