Case study · Healthcare · Datadog HealthScan
Two thirds less log noise, with every warning and error kept
An international eyecare software company cut its daily log volume by about two thirds, and its logs and traces now join up.
At a glance
About two thirds
less daily log volume
More than 90%
log and trace correlation coverage
Every
Datadog best-practice logs check passing
Eyecare softwareAzure.NETDatadog Log ManagementDatadog APMDatadog Professional Services
01
The challenge
The challenge
Datadog makes it easy to index every log a platform sends, and as a platform grows, routine logs bury the warnings and errors engineers need during an incident. When service names come from the infrastructure rather than the application, logs and traces stop joining up and people struggle to find their own service. The company, which runs on Azure, chose to bring in Datadog specialists to get more from the platform it already had.
02
What we did
What we did
- Assessed the Datadog environment and turned it into a prioritised 30/60/90 day plan.
- Redesigned which logs are kept, in code: routine noise dropped or sampled, warnings and errors kept in full, with no logs lost during the switch.
- Validated the team's own tagging standard and joined logs to traces.
- Built monitoring as code, with anomaly detection on the most critical service and runbook guidance inside each alert.
- Made the service catalog filterable by customer, and strengthened data protection in logs.
03
The outcome
The outcome
- Daily log volume down by about two thirds, with every warning and error kept.
- Engineers can follow an issue from logs to traces, with more than 90% correlation coverage.
- Alerts tell the responder what to do, and every monitor is version-controlled.
- Engineers can go straight to one customer's services.
This case study is anonymised, so the customer is not named.
Talk to us about Datadog HealthScan.
Tell us what you run today and what you need from it. We’ll recommend the simplest first step.