How OpenTelemetry unlocked observability flexibility

At CloudAMQP, we've always believed in open standards. We built our service on RabbitMQ, one of the world's most trusted open-source message brokers and have spent nearly a decade running a commercial, hosted service around RabbitMQ. In the spirit of this open-source commitment, a decision was made to switch from our homegrown metrics pipeline to OpenTelemetry, an open-source framework for exports like logs & metrics.

When open-source is not open

We used to build integrations one vendor at a time. If customers wanted Datadog? We built an integration. CloudWatch? We built another. And, if customers were switching tools, that meant starting over. Over time, this approach pushed us to find a better solution. However, it also raised the question: if the greatest power of open source is the freedom to choose what works best - what good is running open-source software if observability data is trapped in proprietary formats and if switching monitoring tools means rewriting instrumentation across all services? This is when OpenTelemetry was brought to the table.

OpenTelemetry: Instrument once, observe anywhere

OpenTelemetry is a unified standard for collecting and exporting metrics, traces, and logs. This "Instrument once, observe anywhere" solution integrates with observability platforms such as Prometheus, Grafana, Datadog, New Relic, and self-hosted solutions. We believe open metrics matter just as much as open source, and with OpenTelemetry, we can provide our customers the same freedom for their observability data as open source provided for their software.

The practical impact

So how does this play out in practice? Our customers can now connect to any OpenTelemetry-compatible platform with standard format and configuration, eliminating the wait for custom integrations. Testing new platforms becomes easier by routing metrics alongside existing tools, without any code changes or risk.

This change also leads to improved cost control since metrics can be filtered and sampled before leaving the network. Only sending high-resolution data where it matters saves money elsewhere. Also, new observability tools will support OpenTelemetry from day one, so customers will never wait for another integration to be built.

There are already a number of integrations leveraging OpenTelemetry available. Try them out on your running cluster or spin up a new cluster and test it out. If you have any questions or considerations about this blog, reach us at contact@cloudamqp.com

Until next time,
CloudAMQP team