Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
teratailの解析基盤をEFKで作っていろいろ楽しい話
Search
ikuwow
March 04, 2016
Technology
920
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
teratailの解析基盤をEFKで作っていろいろ楽しい話
teratailの解析基盤をEFKで作っていろいろ楽しい話 @ ゆとりエンジニア交流会
ikuwow
March 04, 2016
More Decks by ikuwow
See All by ikuwow
Elasticsearch on EC2からAmazon Elasticsearch Serviceに 移行してだいぶ楽になった話
ikuwow
0
3.6k
意外と使える! Alibaba Cloud
ikuwow
0
250
UNIXという考え方
ikuwow
1
2.1k
技術書紹介 パーフェクトPHP
ikuwow
0
2.2k
みんなもMiddlemanで技術ブログ作って幸せになろう!
ikuwow
0
1k
PHPサイバーテロの技法 書籍紹介
ikuwow
0
1k
Other Decks in Technology
See All in Technology
『三匹の子ぶた』から学ぶネットワークセキュリティの昔と今 / Network Security: Then and Now Through the Lens of The Three Little Pigs
nttcom
1
1.8k
トヨタ⽣産⽅式(TPS)⼊⾨
recruitengineers
PRO
3
690
TypeScript入門 2026
recruitengineers
PRO
3
530
オートロックマンションなのに、各部屋は施錠なし!? 攻撃者が組織内ネットワークで大暴れする理由 / The Front Door Is Locked, but the Rooms Are Wide Open: Why Attackers Move Freely Inside Enterprise Networks
nttcom
1
1.6k
社内の7割が使うデータ基盤を、 データチーム2人で回すためにやったこと
koh_yoshi
4
1.3k
AIは実装を速くする。では、私たちは何を今作るべきか?-立場を越えてリリースに向き合ったチーム開発の実践 / 20260801 Hiromi Nakaya and Naoki Takahashi
shift_evolve
PRO
3
470
Issue設計から始める仕様駆動開発 / 20260731 Mizuki Hirata
shift_evolve
PRO
1
150
【CEDEC2026】コードレビュー支援ツール開発から学ぶ:LLMを用いた業務システムの実践的な運用設計と誤出力対策
cygames
PRO
0
640
【AG-UI × A2UI × MCP Apps】Generative UIをやさしく解説する
nrinetcom
PRO
1
110
モノリス Rails でも日中に rails db:migrate を走らせたい! / Daytime rails db:migrate on Monolithic Rails!
euglena1215
4
560
【CEDEC2026】専門性の高いデフォルメチームが挑んだ人材育成戦略 〜Cygames Academiaの企画から実施まで〜
cygames
PRO
0
550
Webアクセシビリティ入門 2026
recruitengineers
PRO
3
490
Featured
See All Featured
Exploring the Power of Turbo Streams & Action Cable | RailsConf2023
kevinliebholz
37
6.5k
The innovator’s Mindset - Leading Through an Era of Exponential Change - McGill University 2025
jdejongh
PRO
1
230
Making the Leap to Tech Lead
cromwellryan
135
10k
Exploring the relationship between traditional SERPs and Gen AI search
raygrieselhuber
PRO
2
4.2k
Automating Front-end Workflow
addyosmani
1369
210k
Understanding Cognitive Biases in Performance Measurement
bluesmoon
32
3k
Lightning talk: Run Django tests with GitHub Actions
sabderemane
0
230
Designing Dashboards & Data Visualisations in Web Apps
destraynor
232
55k
RailsConf & Balkan Ruby 2019: The Past, Present, and Future of Rails at GitHub
eileencodes
141
35k
The Illustrated Guide to Node.js - THAT Conference 2024
reverentgeek
1
420
Building Adaptive Systems
keathley
44
3.2k
Chasing Engaging Ingredients in Design
codingconduct
0
260
Transcript
teratailͷղੳج൫Λ EFKͰ࡞ͬͯ ͍Ζ͍Ζָ͍͠ @ikuwow ϨόϨδʔζגࣜձࣾɹςΫϊϩδʔϝσΟΞϥϘ ΏͱΓੈΤϯδχΞަྲྀձʢ2016/03/04ʣ
ࣗݾհ • ϨόϨδʔζגࣜձࣾɺςΫϊϩδʔϝ σΟΞϥϘɺteratailͷ։ൃͯ͠Δਓɻ • ֶੜͷ࣌εϩʔΨϯגࣜձࣾͰ1.5͙ Β͍Πϯλʔϯͯͨ͠ • ίʔυॻ͘ͱ͖PHPͰ͕͢ɺϑϩϯτ ΠϯϑϥͬͨΓ͍Ζ͍ΖΓ·
͢ • ࠷ۙͬͨ͜ͱɿteratailͷϩάղੳج ൫࡞Δ @ikuwow
teratail ͬͯΔਓʙʁ
teratail • ΤϯδχΞɾϓϩάϥ ϚͷͨΊͷQ&AαΠτ • ຖ࣭͕70-80݅ • ճ93% • 3/17ʹϢʔβʔձʮू
·ͬtailʯୈ࢛ճ։࠵༧ ఆ
ࠓ͢͜ͱ • teratailͷϢʔβʔߦಈϩάΛEFKελοΫ (Elasticsearch, Fluentd, KibanaʣͰՄࢹԽ͢ ΔΈ࡞ͬͨ • ָ͍͠ʂ •
ਏ͍ʂʂ
ϢʔβʔͷߦಈΛݟ͍ͨʂ 1. ϦΞϧλΠϜʹࢹͯ͠ϦεΫݕͨ͠Γɺ ΧδϡΞϧʹ࠷ۙͷϢʔβʔͷಈ͖Λͬͨ Γ͍ͨ͠ʂ 2. KPIΛݟΔͷʹ࠷దԽͨ͠ܗͰσʔλΛ࣋ͬ ͯਂ͘ૣ͘ՄࢹԽ͍ͨ͠ ʢ͋ͱHiveQLॻ͘ͷΊΜͲ͍͍͔͘͢͠͝Βૣ͍ͷʹ͍ͨ͠ɾɾɾʣ
࡞ͬͨج൫ Amazon S3 Amazon Redshift ҹϩάͷྲྀΕ 1. ϦΞϧλΠϜՄࢹԽ 2. ਂ͘ՄࢹԽ
͏গ͚ͩ͠ৄ͘͠ node.master: false node.data: false node.master: true node.data: true node.master:
false node.data: false node.master: true node.data: true Amazon Redshift Amazon S3 teratailͷதͷਓ ४ϦΞϧλΠϜՄࢹԽ KPIΛਂ͘ՄࢹԽ όονॲཧ
Fluentdͱ • ϩάͷύʔεɺूΛ͢Δπʔ ϧ • TreasuredataʢຊͰΘΓͱ ਓؾʣ • Α͘Logstashͱൺֱ͞ΕΔ •
όοϑΝϦϯάݡͯ͘ɺ5͙ Β͍ࢭΊͯશ͘ͳ͍
Elasticsearchͱ • ࠷ۙྲྀߦΓͷશจݕࡧΤϯδϯɻ2ܥ ͕࠷৽ɻ • ElasticࣾʢLogstashͱಉ͡ʣ • ͖Ε͍ʹRESTfulͳAPIͰѻ͍͍͢ • ͱΓ͋͑ͣಉ͡ωοτϫʔΫʹஔ͍
͓͚ͯΫϥελ࡞ͬͯ͘ΕΔ • ࠷ۙAWS͕Elasticsearch Serviceͱ ͍͏ͷΛग़ͨ͠Γ
Kibanaͱ • ElasticsearchΛόοΫͱ͠ ͯɺͦΕΒͷσʔλΛ͔ͬ ͜ྑ͘ՄࢹԽ͢Δπʔϧ • nodeΞϓϦέʔγϣϯͳͷ Ͱಋೖָ͕͘͢͝ • ϚεϙνϙνͰϩά͕ݟΒ
ΕΔ
EFKελοΫͷಛ • Πϯετʔϧཧ͕ൺֱతΧϯλϯ • FluentdϫϯϥΠφʔ͚ͩͰ͍͚Δ • Elasticsearchউखʹ͏·͍͜ͱΫϥελ࡞ͬͯ͘ΕΔ • KibanaೖΕΔͷ؆୯ͩ͠ݟͨΒ͍͍ͩͨ͑Δ •
ͦͦ͜͜ރΕ͖ͯͨײ͋Δʁ • ࢼͯ͠ΈΔͱ͍͕͙͢͢͞
࡞ͬͯԿ͕มΘ͔ͬͨʁ • ϩά͕؆୯ʹૣ͔ͬ͘͜Α͘ݟΒΕΔ༷ʹͳͬͨ • ࣌ؒͷॖ • νʔϜશһʹɺ͍ܰؾ࣋ͪͰ͍͍͢͢ϩάΛूܭɾՄࢹԽɾੳ͢ Δश׳͕͍ͭͯɺΠϕϯτࣄͷͨͼʹߦಈྔ૿͑ͨΓ͢Δͷ͕Έͯ ָ͍͠ •
ϩάʹײҠೖͰ͖ΔΑ͏ʹͳͬͨʂ • ͓͍߹Θͤ࣌ʹࠔͬͯΔϢʔβʔͷߦಈΛ͑ΔΑ͏ʹͳͬͨ • όάͷݪҼ͕ɺϩά͔ΒϢʔβʔͷಈ͖Λ࠶ݱͯ͠ΈͨΒ໌ͨ͠
ָ͍͠ʂ
΄͔ʹΓ͍ͨ͜ͱ • ApacheͷΤϥʔϩάɺΞΫηεϩάͷՄࢹԽɾੳ • fluentdͰTemplate͕༻ҙ͞Ε͍ͯΔͷͰ؆୯ʹͰ͖Δ • ϨεϙϯελΠϜͱ͔ग़͓ͯ͘͠ͱͬͱָ͍͠ • ΞϓϦέʔγϣϯϑϨʔϜϫʔΫͷΤϥʔϩά •
Fluentdෳߦϩά͍͚Δ • slow queryͷϩάݟͯΨϯΨϯѱ͍ΫΤϦΛ௵͢ ϦΞϧλΠϜੑ͕ٻΊΒΕΔใΛݟ͍͔͢Β͘͢͝Ԡ༻ར͘
ਏ͔ͬͨ͜ͱ • HadoopʹೖΕ͍ͯͨಠࣗͷϑΥʔϚοτΛਖ਼ن දݱͰॻ͘ͷͭΒ͍ • ϩά͕1.3%͙Β͍ܽଛ͢Δ => ࣏ͬͨ • Index
template͚ͭͨΒಡΊͳ͍ͬͯݴΘΕΔ • Autoscaling͕ݡ͗ͯͬͯͨ͢ͷterminate͞Εͨ
<source> @type tail path /home/ikuo.degawa/hogehoge.logs pos_file /tmp/hogehoge.logs.pos format /^(?<dt>[^\t]+)\t(?<site_id>[^\t]*)\t(?<action>[^\t]*)\t(? <option>[^\t]*)\t(?<user_id>[^\t]*)\t(?<session_cookie>[^\t]*)\t(?
<storage_cookie>[^\t]*)\t(?<view_type>[^\t]*)\t(?<user_agent>[^\t]*)\t(? <page_id>[^\t]*)\t(?<url>[^\t]*)\t(?<time>[^\t]*)\t(?<ip>[^\t]*)\t(? <segment>[^\t]*)\t(?<var>[^\t]*)\t(?<view>[^\t]*)\t(?<act>[^\t]*)\t(?<post0>[^ \u0001]*)\u0001(?<post1>[^\u0001]*)\u0001(?<post2>[^\t]*)\t(?<search0>[^ \u0001]*)\u0001(?<search1>[^\u0001]*)\u0001(?<search2>[^\u0001]*)\u0001(? <search3>[^\u0001]*)\u0001(?<search4>[^\u0001]*)\u0001(?<search5>[^\u0001]*) \u0001(?<search6>[^\u0001]*)\u0001(?<search7>[^\t]*)\t(?<user0>[^\u0001]*) \u0001(?<user1>[^\u0001]*)\u0001(?<user2>[^\u0001]*)\u0001(?<user3>[^\t]*)\t(? <other0>[^\u0001]*)\u0001(?<other1>[^\u0001]*)\u0001(?<other2>.*)$/ tag mogmog-logs.gerogero </source> HadoopʹೖΕ͍ͯͨಠࣗͷϑΥʔ ϚοτΛਖ਼نදݱͰॻ͘ͷͭΒ͍
ϩά͕1.3%͙Β͍ܽଛ͢Δ => ࣏ͬͨ • Kibanaͷ݅ͱɺcat hoge.log | wc -l ͨ݁͠Ռ
͕ҧ͏ʂʂ • lotateͨ͠ઌͷϑΝΠϧΛ ಡΈ࢝ΊΔλΠϛϯά͕ ͍ͱ͍͏༷Λൃݟ • read_from_headΛͬͨ Β࣏ͬͨ લͷ ࣍ͷ ͜ͷล͔ΒಡΜͰͨ
Index template͚ͭͨΒಡΊͳ ͍ͬͯݴΘΕΔ • index template: elasticsearchʹೖΔ ࣌ͷmappingΛࢦ ఆͰ͖Δ •
index໊Λ݅ʹܕ ΛܾΊΒΕΔ { "templates": “awesomelog-*", "settings": { "number_of_shards" : 1 }, "mappings": { "awesomelogs" : { "properties" : { "@timestamp" : { "type" : "date", "format" : "strict_date_optional_time||epoch_millis" }, "act0" : { "type" : "integer" }, "act1" : { "type" : "integer" }, "act10" : { "type" : "string", "index": "not_analyzed" }, "act11" : { "type" : "string" }, "act2" : { "type" : "integer" }, "act3" : { "type" : "integer" }, "act4" : { "type" : "string" }, "act5" : { "type": "multi_field", "fields": {
ύϑΥʔϚϯε্͕Δͱࢥͬͨ Βɾɾɾ { "templates": “awesomelog-*", "settings": { "number_of_shards" : 1
}, "mappings": { "awesomelogs" : { "properties" : { "@timestamp" : { "type" : "date", "format" : "strict_date_optional_time||epoch_millis" }, "act0" : { "type" : "integer" }, "act1" : { "type" : "integer" }, "act10" : { "type" : "string", "index": "not_analyzed" }, "act11" : { "type" : "string" }, "act2" : { "type" : "integer" }, "act3" : { "type" : "integer" }, "act4" : { "type" : "string" }, "act5" : { "type": "multi_field", "fields": { • ࣮intΛظ͍ͯ͠Δͱ͜ ΖʹstringඈΜͰ͖ͨΓ͠ ͯͨʢϩάͷ࣮ϛεʣ • ϩά͕ೖͬͨͱ͖ʹΤϥʔ ు͍ͯͯɺfluentdͷόο ϑΝʹཷ·Γଓ͚ͯͨ • ݁ہnot_analyzedΛ͚ͭͨ ͷΈ
Autoscaling͕ݡ͗ͯ͢terminate ͞Εͨ ʂʁ
ʮavailability zone͕Ճ͞Ε͔ͨΒɺόϥϯε Αͯ͘͠Մ༻ੑ͋͛ΔͨΊʹ͍ͬ͜ফͯ࣍͠ͷ ݐͯΔΑʂʯ
ڭ܇ɾɾɾ • Fluentd͓ੈগͳͯ͘ࡁΉ͕ɺϩάͷಡΈ ํΛͬͱ͚ • ElasticsearchElasticʹ͓͍ͯͨ͠΄͏͕͍͍ • Auto Scaling Groupݡ͍
·ͱΊ • KibanaͰϩάΛ͔ͬ͜Α͘ݟΒΕΔͱσʔλ ʹײҠೖͰ͖ΔΑ͏ʹͳΓɺνʔϜશһ͕ ϢʔβʔͷߦಈΛݟΒΕΔਓʹͳΕΔ • ָ͍͠
ฐࣾͰΤϯδχΞΛืूதͰ͢ ͝ਗ਼ௌ͋Γ͕ͱ͏͍͟͝·ͨ͠
͜ͷຊʹ͓ੈʹͳΓ·ͨ͠ • ͍͍ຊͰ͢
@ikuwow