Ask HN: Do you use OpenTelemetry in your organisation?

[siddharthgoel88]

Hello folks!

I have noticed that in past few years the topic of observability is becoming more and more popular. And in particular OpenTelemetry is something that sees similar increasing interest ( https://trends.google.com/trends/explore?date=today%205-y&q=OpenTelemetry,Observability&hl=en ) .

With these, some questions come to my mind and I wish to learn from the HN community about their view on this.

  - Do you use or plan to use OpenTelemetry in your organisation? Why or why not?
  - If you use, then how do you use it? May be a bit about architecture and vendors you rely on.
  - I think currently no vendor is able to do justice to the view for the OTel data collected. I mean with OTel conceptually merges the signal Metrics, Logs and Traces. But the vendors have not unified a view for it yet (as what I have seen). Do you agree? If yes, then how do you tackle this?

Looking forward to learning for the community here!

[EspressoGPT]

In my previous job, I've integrated OpenTracing and OpenTelemetry into our microservices. You need to do this in microservice infrastructures to be able to debug and investigate service calls. We first used Jaeger for tracing calls across service boundaries, and then migrated to Instana.

Instana is a wholistic solution and does a pretty good job merging and displaying the collected metrics, logs, and traces. I did prefer Jaeger for tracing simple service calls because it isn't as complex as Instana from a developer perspective, but if you need a deep insight into your running servers, operating systems, runtimes, containers, and applications and how they interact with each other, Instana is really powerful.

[cratermoon]

I helped set it up. We gathered usage stats for various features in addition to the primary use, monitoring and alerting. We have a web front end calling backend services with grpc, and inter-service calls with grpc. One of the key needs was being able to trace calls from the front end all the way to backend, across multiple services in a heterogeneous kubernetes cluster. We fed the data into Sumo Logic and mostly used the built-in dashboards, monitoring, and alerting. The was little call for anything more sophisticated, as the organization was somewhat immature in the SRE department. My impression of Sumo Logic was very favorable and it might fill many of your use cases.

[siddharthgoel88 → cratermoon]

Thanks for your response! I will look into Sumo Logic in that case :-)