Masahiro Nakagawa 
Nov 6, 2014 
Cloudera World Tokyo 
Fluentd 
Unified Logging Layer
Who are you? 
> Masahiro Nakagawa 
> github/twitter: @repeatedly 
> Treasure Data, Inc. 
> Senior Software Engineer 
> Fluentd / td-agent developer 
> I love OSS :) 
> D language - Phobos committer 
> Fluentd - Main maintainer 
> MessagePack / RPC- D and Python (only RPC) 
> The organizer of Presto Source Code Reading 
> etc…
Structured logging 
! 
Reliable forwarding 
! 
Pluggable architecture 
http://fluentd.org/
Data Processing 
Data source 
Collect Store Process Visualize 
Reporting 
Monitoring
Related Products 
Collect Store Process 
Visualize 
Cloudera 
Horton Works 
Treasure Data 
Excel 
Tableau 
R 
easier & shorter time 
???
What’s Fluentd? 
• Open sourced log collector written in Ruby 
• Customization is essential 
small core + many plugins 
6 
It’s like syslogd, but 
uses JSON for log messages
Divide & Conquer & Retry 
7 
error retry 
retry 
error retry retry
Core Plugins 
! 
• Divide & Conquer 
• Buffring & Retrying 
• Error handling 
• Message routing 
• Parallelism 
! 
! 
• read / receive data 
• write / send data 
8
Pluggable Architecture 
9 
Pluggable Pluggable 
Buffer Output 
Input 
> Forward 
> HTTP 
> File tail 
> dstat 
> ... 
> Forward 
> File 
> MongoDB 
> ... 
> File 
> Memory 
Engine 
Output 
> rewrite 
> ...
Before 
10
After 
11
M x N → M + N 
12 
Alerting 
Nagios 
Analysis 
MongoDB 
MySQL 
Hadoop 
Archiving 
Amazon S3 
Access logs 
Apache 
App logs 
Frontend 
Backend 
System logs 
syslogd 
Databases 
buffer / buffer / routing
Ecosystem
Treasure Agent 
• Treasure Data distribution of Fluentd 
• including Ruby, core libraries and 
third party plugins 
• Current stable is 2.1.1 
• rpm / deb / dmg 
• 2.1.2 will be released soon with fluentd-ui 
14
fluentd-forwarder 
• Forwarding agent written in Go 
• mainly for Windows support 
• immature than Fluentd 
• Bundle TCP input/output and TD output 
• No plugin mechanizm 
• Similar product: ik, fluent-agent-hydra, etc 
15
fluentd-ui 
• Manage Fluentd instance using Web GUI 
16
Use-case
Users 
• Companies 
• Web service, Ad Tech, Game, Media, 
EC, IoT, PoS, etc.. 
• Products 
• Kubernetes, Cloudn, Harvester, 
Bloomberg Clustered Private Cloud, etc.. 
18
Simple forwarding 
19 
HDFS 
HBase 
Mongo 
S3
Separate log stream 
20
CEP for stream processing 
21
Roadmap 
• v0.10 (current) 
• v0.12 (Nov, 2014) 
• Filter / Label / at-least-once 
• v0.14 (TBD) 
• New Plugin APIs, ServerEngine 
• v1 (TBD. In the first half of 2015) 
22
Cloud service for the entire data pipeline 
Check: treasuredata.com

Fluentd: Unified Logging Layer at CWT2014

  • 1.
    Masahiro Nakagawa Nov6, 2014 Cloudera World Tokyo Fluentd Unified Logging Layer
  • 2.
    Who are you? > Masahiro Nakagawa > github/twitter: @repeatedly > Treasure Data, Inc. > Senior Software Engineer > Fluentd / td-agent developer > I love OSS :) > D language - Phobos committer > Fluentd - Main maintainer > MessagePack / RPC- D and Python (only RPC) > The organizer of Presto Source Code Reading > etc…
  • 3.
    Structured logging ! Reliable forwarding ! Pluggable architecture http://fluentd.org/
  • 4.
    Data Processing Datasource Collect Store Process Visualize Reporting Monitoring
  • 5.
    Related Products CollectStore Process Visualize Cloudera Horton Works Treasure Data Excel Tableau R easier & shorter time ???
  • 6.
    What’s Fluentd? •Open sourced log collector written in Ruby • Customization is essential small core + many plugins 6 It’s like syslogd, but uses JSON for log messages
  • 7.
    Divide & Conquer& Retry 7 error retry retry error retry retry
  • 8.
    Core Plugins ! • Divide & Conquer • Buffring & Retrying • Error handling • Message routing • Parallelism ! ! • read / receive data • write / send data 8
  • 9.
    Pluggable Architecture 9 Pluggable Pluggable Buffer Output Input > Forward > HTTP > File tail > dstat > ... > Forward > File > MongoDB > ... > File > Memory Engine Output > rewrite > ...
  • 10.
  • 11.
  • 12.
    M x N→ M + N 12 Alerting Nagios Analysis MongoDB MySQL Hadoop Archiving Amazon S3 Access logs Apache App logs Frontend Backend System logs syslogd Databases buffer / buffer / routing
  • 13.
  • 14.
    Treasure Agent •Treasure Data distribution of Fluentd • including Ruby, core libraries and third party plugins • Current stable is 2.1.1 • rpm / deb / dmg • 2.1.2 will be released soon with fluentd-ui 14
  • 15.
    fluentd-forwarder • Forwardingagent written in Go • mainly for Windows support • immature than Fluentd • Bundle TCP input/output and TD output • No plugin mechanizm • Similar product: ik, fluent-agent-hydra, etc 15
  • 16.
    fluentd-ui • ManageFluentd instance using Web GUI 16
  • 17.
  • 18.
    Users • Companies • Web service, Ad Tech, Game, Media, EC, IoT, PoS, etc.. • Products • Kubernetes, Cloudn, Harvester, Bloomberg Clustered Private Cloud, etc.. 18
  • 19.
    Simple forwarding 19 HDFS HBase Mongo S3
  • 20.
  • 21.
    CEP for streamprocessing 21
  • 22.
    Roadmap • v0.10(current) • v0.12 (Nov, 2014) • Filter / Label / at-least-once • v0.14 (TBD) • New Plugin APIs, ServerEngine • v1 (TBD. In the first half of 2015) 22
  • 23.
    Cloud service forthe entire data pipeline Check: treasuredata.com