Data Pipeline - Sub traces

Elastic Observability APM
1

Hey!

We are integrating elastic APM in our data pipeline, using the latest python APM client.
There are several components that analyze an item one after another.

Each component (microservice) receives traceparent and tracestate information from its parents, and begins a transaction.

Currently, we can only see one trace in the system, which is the first component in the pipeline.
image

Inside the trace, we can see the sub components operations, but there are no statistics available about them.
image

We would like to receive better statistics on the total time of the trace, including all child transactions

Are there ways to achieve this?

Thanks!

2

Hi @Koby_Bass !
Could you add more details to your question please?
I want to make sure I understand what you mean by those statistics.
Thanks!

3

We would like to see:
What's the latency between the start of service_one until the end of service_three?
What's the throughput of the entire system, starting at service_one to service_three?

Ideally, instead of seeing a single trace in the trace screen, I would like to see 4:
one for each service separately - service_one, two and three
one for the system as a whole

Does that help?

4

Got it.
Normally we would show exactly that WHEN you have an overlapping parent transaction, like in the example below.

image
image3884×1006 187 KB

How do services call each other?
Is there a parent pipeline orchestration which could be instrumented end-to-end?

5

This topic was automatically closed 20 days after the last reply. New replies are no longer allowed.