Header background

Breaking data silos: Liquid Reply’s journey to custom API observability with OpenTelemetry and Dynatrace

Modern IT organizations are generating more data from more tools and technologies than ever. Data is proliferating in separate silos from containers and Kubernetes to open source APIs and software to serverless compute services, such as AWS and Azure. Moreover, distributed microservices architecture and data silos mean teams don’t have access to the context that’s critical to make sense of this data and all its different data formats. As a result, teams are in dire need of end-to-end observability that can accommodate disparate formats and custom APIs with the context needed to take meaningful action.

The tracing challenge

A large Italian bank group was facing inconsistent data and tracing issues. The organization needed to ensure the correlation of all events in a complete end-to-end trace. However, they had numerous custom applications with separate APIs. Although the APIs were all managed by the Google API manager Apigee, the bank group was not getting consistent data types from the output.

Additionally, due to performance impact, Apigee’s distributed tracing only configures a probabilistic sampling rate, with a maximum rate of 50%. This restriction made it impossible to totally export all traces from Apigee unless the sampling rate was below the maximum, in which case they could use Google Cloud Trace. In this case, the bank group could not rely solely on Google Cloud Trace because they needed to collect traces and monitor the applications across all their systems. For this comprehensive monitoring in context, they had adopted Dynatrace. They also wanted to normalize trace data from their custom APIs to simplify ingestion and provide complete end-to-end visibility.

Solving custom API observability with OpenTelemetry

Because the bank group needed complete coverage for all API traffic, it wanted to standardize tracing on all API calls. In addition to being Dynatrace users, the bank group also enlisted the help of Liquid Reply, a cloud services company and Dynatrace partner specializing in container orchestration, cloud-native development, and FinOps.

By leveraging Liquid Reply’s knowledge of OpenTelemetry and Dynatrace’s contribution to the open-source standard for collecting observability data, the bank group ultimately achieved 100% coverage of traces. Liquid Reply was able to normalize custom APIs to make them ready to ingest for end-to-end tracing by Dynatrace.

The Google Cloud Community article, Sending OpenTelemetry spans from Apigee hybrid to Dynatrace, describes how to send all API traces to Dynatrace by leveraging OpenTelemetry open standards and without enabling the native distributed tracing option in Apigee hybrid.

The approach described in the article uses the OpenTelemetry Collector to convert the OTLP protocol in JSON format to the OTLP/HTTP format. The first step allowed Liquid Reply to get the data needed for monitoring in a format that’s fully manipulable on the source-side thanks to Apigee policies capabilities. The process also preserves trace context data so Dynatrace can correlate traces from their origin to the edge. By adequately configuring the OpenTelemetry Collector, Liquid Reply could seamlessly ingest OpenTelemetry traces into Dynatrace, resulting in REST APIs that were properly formatted for constructing end-to-end traces using PurePath®, the patented distributed tracing and code-level analysis technology from Dynatrace.

Liquid Reply: OpenTelemetry + Dynatrace = Better together

Liquid Reply has learned that integrating OpenTelemetry with Dynatrace is a seamless and straightforward process. Dynatrace provides everything from application and infrastructure monitoring to digital experience and security, leveraging data in context, AI, and intelligent automation across the full stack for end-to-end observability. When it comes to custom situations like the case with Liquid Reply’s large banking group client, OpenTelemetry and Dynatrace are better together.

In this instance, Liquid Reply recognized that using OpenTelemetry was the right approach for instrumenting the banking group’s custom APIs for easy ingestion into Dynatrace for end-to-end tracing. Dynatrace’s flexibility in features, openness to external data ingestions, and contributions to the OpenTelemetry Project enabled a path forward in this challenging situation. Together, partners Dynatrace and Liquid Reply can overcome technological limitations to help customers reach full observability.