Fluent Bit Log Collection

⬢ 티어 2기술

높음

연봉 영향

2개월

학습 기간

보통

난이도

—

직무

한눈에 보기

Fluent Bit is a lightweight, fast log aggregator written in C. It's 100x smaller than Fluentd (used in Kubernetes, IoT, edge computing). Companies like Datadog, Splunk, and AWS use Fluent Bit as the backbone of observability pipelines. Professionals earn 85-95k USD junior, 150-180k senior. Learning the config, plugins, and filtering takes 4-6 weeks. Sits between raw log files (chaotic) and centralized observability platforms (Datadog, ELK). The scarcity: most engineers default to Fluentd or Logstash; Fluent Bit power users are 5% of the market.

Fluent Bit Log Collection이란

Fluent Bit is a lightweight, high-performance log collector and shipper written in C. It reads logs from files, sockets, stdin, systemd journals, and container outputs, parses them (JSON, regex, multiline), filters noisy events, and forwards to centralized systems like Elasticsearch, Datadog, Splunk, or AWS CloudWatch. Fluent Bit is 100x smaller than Fluentd (1MB vs 100MB) and uses 10x less CPU. It's the standard for edge devices, IoT, Lambda functions, and containerized workloads. Every Kubernetes cluster running observability needs Fluent Bit (or similar) as the log collection layer.

🔧 도구 & 생태계

Fluent Bit C libraryFluent Bit config language (HOCON)Input plugins (stdin, TCP, syslog, files, systemd)Output plugins (Elasticsearch, Datadog, Splunk, S3)Filter plugins (regex, parser, modify)Lua scripting (custom filters)Docker logging driverKubernetes DaemonSet deploymentPrometheus metrics (Fluent Bit telemetry)Parser libraries (JSON, regex, multiline)

💰 지역별 연봉

지역	주니어	미들	시니어
USA	$86k	$148k	$215k
UK	$52k	$90k	$130k
EU	$56k	$95k	$140k
CANADA	$83k	$142k	$205k

🎓 자격증

Fluent Bit Official Documentation Fluent Project Slack and Community Linux Foundation LF Training (LFCA)

❓ 자주 묻는 질문

When should you use Fluent Bit vs Fluentd?

Fluent Bit for edge devices, containers, low-memory environments (IoT, Lambda, ECS). It's 100x smaller, uses 10x less CPU. Fluentd for on-premise servers and large-scale log processing. Fluent Bit typically feeds into a central Fluentd or goes directly to Elasticsearch/Datadog.

How do you parse multiline logs (stacktraces) with Fluent Bit?

Use the multiline parser and specify regex patterns for line starts (e.g., timestamp pattern for new log entry). Stack traces that span multiple lines are joined until the next log entry starts. Test parsers with fluent-bit -c config.conf -dry-run.

Can Fluent Bit filter or drop logs?

Yes. Use filter plugins to regex-match and drop noisy logs. Example: exclude health checks (GET /health 200), drop debug logs below WARNING level. Filtering at source saves storage and reduces Elasticsearch cardinality explosion.

How do you deploy Fluent Bit in Kubernetes?

Deploy as a DaemonSet (one pod per node). Mount /var/log and /var/lib/docker/containers to read host logs and container logs. Configure it to parse JSON-formatted logs from containers. Use Prometheus ServiceMonitor to scrape Fluent Bit's /api/v1/metrics endpoint.

What are the performance limits of Fluent Bit?

Fluent Bit can handle 100k+ events/sec on modern hardware. Bottleneck is usually the output (network, Elasticsearch throughput). Use buffering and batching to avoid overwhelming downstream systems. Monitor via Prometheus metrics (input records/sec, output errors).