Case study

Industry
Security and alarm services
Platforms
Self-managed Elastic, Elastic APM, OpenTelemetry, Kubernetes, Helm, Ansible
Duration
Since Q1 2026, ongoing
signals connected: traces, logs and metrics
investigation surface, no tool switching
big-bang cutover
The challenge
A European security-sector organization runs distributed services across on-premises and cloud platforms. It already produced plenty of telemetry, but investigations relied on logs: operators rebuilt request journeys by hand, across several tools.
The gap was context, not volume. Inconsistent service names and metadata made logs, metrics and traces hard to connect during incidents. Production had to stay stable throughout, so a big-bang cutover was not an option.
What we found
A review of the architecture, validated in a non-production environment, showed the main gap was in correlation and service identity, not in tooling. OpenTelemetry data was already arriving alongside the existing APM path.
What we did
Designed the end-to-end observability architecture, from workloads to collectors, gateway, Elasticsearch and Kibana.
Defined a telemetry contract: consistent service names, identity fields and trace-ID correlation, aligned with Elastic Common Schema.
Designed the OpenTelemetry ingestion path, running in parallel with the existing APM and Beats, so investigation capability never dropped.
Set lifecycle, sampling and cardinality governance for the hybrid phase.
Sequenced the next steps: buffering, tail sampling, wider adoption and service-level objectives.
Results
OpenTelemetry ingestion established on Elastic, alongside the existing APM path.
One investigation surface: trace IDs and consistent service identity link traces to related logs, without switching tools.
Clear governance for a progressive migration: identity, mappings, lifecycle and sampling are defined.
What it means for the client
Teams can start trace-based investigation now, without waiting for a full cutover, and have a clear path to deepen adoption. Expected next: faster root-cause analysis as more services adopt OpenTelemetry, clearer service maps, and service-level objectives once the telemetry contract is in place.

