Fast logs
summary
Shows that deferring expensive formatting with a lazy logger (pass a Supplier) avoids calling a 5s getPrice and drops cost to ~1ms, that most frameworks (Logback, Log4j2) support this, and that switching to an async appender and tuning buffer size trades throughput vs potential message loss. Replacing Logstash by emitting JSON (Logback -> filebeat -> Elasticsearch -> Kibana) removes a hop and speeds ingestion but requires proper JSON escaping; alternatively send events to Kafka or a Docker logging driver for durable business metrics. Schema-on-write yields faster reads.
AI-written summary · about
chapters (20)
- 0:00Start
- 0:48Hazelcast
- 1:24The root of all evil
- 6:25We are bound to the physical world...
- 7:22Writing the stream
- 7:59Synchronous vs. asynchronous logging
- 8:45A sample configuration
- 9:04Example: Logback configuration
- 10:35Associated meta-data
- 11:02Some metadata is expensive to get
- 11:43Logs aggregation
- 12:13Additional metadata
- 12:43Searching in logs
- 14:01Example architecture: the Elastic Stack
- 15:30Sample output
- 17:35Food for thoughts
- 19:47Logs are not only technical
- 20:23Logging options
- 23:30Summary
- 24:48Takeaway #2
