Abstract (EN):
Logging has long been a pillar for monitoring and troubleshooting software systems. From server and infrastructure to application-specific data, logs are an easy and quick way to collect information that may prove useful in diagnosing future issues. When systems become distributed, as is common on the cloud, logs are harder to collect and process. This paper presents three design patterns for logging in cloud-native applications. Standard Logging advises using a standard format for logs across all services and teams so they are easier to process by humans and machines. Audit Logging suggests that important user actions and system changes are recorded in a data store to ensure regulatory compliance or help investigate user-reported issues. Lastly, Log Sampling is about prioritizing logs to maintain a manageable amount of storage. These patterns were mined from existing literature on logging and cloud best practices to make them simpler to communicate, more detailed, and easier for all practitioners to understand.
Language:
English
Type (Professor's evaluation):
Scientific