Use the dependency map
See who calls whom according to the traces, with each service's health and each connection's traffic, to find where a problem originates.
The map draws your services and what they use (databases, external APIs, queues and RPC) based on real traces. The most-called items sit in the center, with what uses them around them.
Before you start
View permission in the APM module.
Instrumented services with traffic in the period.
Steps
Go to APM → Dependency map.
Choose the environment in All environments (for example, production only).
Choose the period: 15 min, 1 h, 6 h, 24 h or 7 days.
Choose the view:
Network: a graph with the services and their connections.
Layers: columns from left to right according to who calls whom, with zoom (−, + and the percentage to fit to width).
Hover over a service or a connection to highlight its path and see its traffic, errors and latency.
Click a service to open its detail page.
How to read it
Element | Meaning |
|---|---|
Green | Healthy |
Yellow | Slow or no data |
Red | With errors |
Gray | No traffic |
Line thickness | How much traffic goes through that connection |
Databases, external APIs, queues and RPC come from the outgoing calls your services record. Connections between two of your own services are calculated by matching the parts of the same trace.
If a red service calls a database that is also red, the problem is most likely in the database. Open the service's detail page, Dependencies tab, to confirm the error rate and p95 of that call.
Show inactive services
Services without telemetry in the last few days are not shown. If any are hidden, check Show inactive to see them.