Pattern: Metrics over Logs

Use metrics as the primary data source to monitor application and system health

Metrics over Logs

A team is moving from a legacy style monolithic system to a cloud native platform. Traditionally most information from applications gets pushed into logs. Cloud Native systems offer a better approach to gaining insight into our applications.

In This Context

When building applications, traditionally we send all relevant information out as logs. Monitoring or metrics were mainly used for the health of our physical infrastructure. Applying this pattern to a modern distributed system composed of dozens of components results in a petabytes of mostly useless data.


The purpose of each type of data, logs and metrics, must first be well established. Logs should be used as a secondary method for gaining further knowledge on a specific incident (i.e. debugging). Metrics are a way to get a holistic view on your applications and system as a whole.


The team can get a clear, centralized view on the overall health of the system and its components.