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
データ基盤を作るのに使うようなサービスは単発で使うだけでも便利 / Manage to loo...
Search
polamjag
January 23, 2019
Programming
210
1
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
データ基盤を作るのに使うようなサービスは単発で使うだけでも便利 / Manage to look over unmanaged data with fully managed services
ref.
http://developer.hatenastaff.com/entry/2019/01/11/110000
polamjag
January 23, 2019
More Decks by polamjag
See All by polamjag
転ばぬ先のXS入門 / YAPC::Fukuoka 2025
polamjag
0
570
障害対応を起点としたもっといい開発と運用のサイクル作りのためにできること / Hatena Enginner Seminar #29
polamjag
0
1.1k
GitHub Linguistと学ぶ!全国このコードはPerlとRakuどっちでしょうクイズ選手権大会 in YAPC::Hiroshima 2024 #yapcjapan
polamjag
0
900
はてなブログとチーム構成とスクラムのこの1年 #dmm_hatena
polamjag
0
3.4k
Google I/O で見た Android Q 以外の話 / #io19 recap
polamjag
0
180
Other Decks in Programming
See All in Programming
AI時代のPHPer生存戦略 ~「言語、もうなんでもよくない?」に本気で向き合う~
vivion
0
220
AI時代に設計が 最大の生産性レバーになる 意図駆動開発とデータを消さない設計|Don't Delete Your Data or Your Intent — Design as the Deepest Lever in the AI Era
tomohisa
0
450
Foundation Models frameworkで画像分析
ryodeveloper
1
430
freee が目指す データ マネジメント戦略 AI-Ready 時代を支える 攻めのガバナンスとは
freee
PRO
0
120
Embedded SREと共に達成した会員管理システムのAWS移行 - SRE NEXT 2026 ランチスポンサーセッション
niftycorp
PRO
1
3.3k
torikago - Ruby::Boxで照らすモジュラモノリスの実行境界
se4weed
1
320
型も通る、synthも通る、それでも危ない 〜AIのCDKの権限とコストを機械で検証する〜 / It Passes Type Checks, It Passes Synth Checks, but It’s Still Risky — Automatically Verifying Permissions and Costs in AI’s CDK —
seike460
PRO
1
520
生成AIで帳票OCRが「簡単に」作れる時代になった?
kon_shou
0
300
PHP初心者セッション2026 〜生成AIでは見えない裏側を知る:今だからLAMPを通して仕組みを学ぶ〜
kashioka
0
780
php-fpmのプロセスが枯渇した日-調査・対処・そして本当にやるべきだったこと-
shibuchaaaan
0
190
Japan Community Day at Kubecon + CloudNativeCon Japan 2026: Learning Container Privilege Control by Building My Own Low-Level Container Runtime
ternbusty
1
120
PostgreSQL 18で考えるUUID主キー
kazuhiro1982
0
450
Featured
See All Featured
Pawsitive SEO: Lessons from My Dog (and Many Mistakes) on Thriving as a Consultant in the Age of AI
davidcarrasco
0
200
Fashionably flexible responsive web design (full day workshop)
malarkey
408
67k
Why Your Marketing Sucks and What You Can Do About It - Sophie Logan
marketingsoph
0
340
Reality Check: Gamification 10 Years Later
codingconduct
0
2.2k
Efficient Content Optimization with Google Search Console & Apps Script
katarinadahlin
PRO
1
750
The Hidden Cost of Media on the Web [PixelPalooza 2025]
tammyeverts
2
420
Embracing the Ebb and Flow
colly
88
5.1k
The Spectacular Lies of Maps
axbom
PRO
1
880
A Guide to Academic Writing Using Generative AI - A Workshop
ks91
PRO
1
360
SERP Conf. Vienna - Web Accessibility: Optimizing for Inclusivity and SEO
sarafernandez
2
1.5k
It's Worth the Effort
3n
188
29k
Believing is Seeing
oripsolob
1
180
Transcript
σʔλج൫Λ࡞Δͷʹ͏Α͏ͳαʔϏε ୯ൃͰ͏͚ͩͰศར άϩʔεϋοΫͷจ຺Λఴ͑ͯ Hatena Engineer Seminar #11 2019-01-23 @ Tokyo
/ 2019-01-30 @ Kyoto id:polamjag
whoami • id:polamjag ϙ ϥ Ϝ δ ϟ ά •
ͯͳϒϩά • ΞϓϦέʔγϣϯΤϯδχΞ
ࠓ͢͜ͱ • େ౷Ұσʔλج൫ͷΑ͏ͳଘࡏ͕ແ͍ঢ়ଶͰɺ σʔλج൫Λ࡞Δͷʹ͏Α͏ͳαʔϏε ΞυϗοΫͳੳʹ୯ൃͰ͑Δ͚ͩͰศར • Google BigQuery ͱ Amazon
Athena Ͱ ·ͣઌͷ͔Βղܾ
ࠓ͞ͳ͍͜ͱ • ͍͢͝σʔλج൫Λθϩ͔Β࡞͍ͬͯ͘ํ๏ • ͍͢͝σʔλج൫Λ׆༻͢Δํ๏ • ࣮ࡍͷࢪࡦͦͷͷʹ͍ͭͯ • ࣈͷݟํɺଊ͑ํ
άϩʔεϋοΫ https://trends.google.com/trends/explore?date=all&geo=JP&q=άϩʔεϋοΫ
ͱ͍͑ࣈ
(ผʹάϩʔεϋοΫʹݶΒͣ) ͪΐͬͱ͜͏͍͏ࣈΛ ݟͯΈ͍ͨΜ͚ͩͲɺͱ͍͏ͱ͖
(ผʹάϩʔεϋοΫʹݶΒͣ) ͪΐͬͱ͜͏͍͏ࣈΛ ݟͯΈ͍ͨΜ͚ͩͲɺͱ͍͏ͱ͖ αʔϏεͷσʔλϕʔε͔Βूܭ ϩάϑΝΠϧ܈͔Βूܭ
αʔϏεͷσʔλϕʔε͔Βूܭ ϩάϑΝΠϧ܈͔Βूܭ (ͯͳϒϩά MySQL ͳͷͰͦΕΛલఏʹ)
• SQL Λී௨ʹୟ͘ • ॻࣺͯεΫϦϓτΛॻ͍ͯूܭͨ͠ͷΛεϓϨο υγʔτʹΞοϓϩʔυ • ͦ͏͍͏ͷΛόονͱͯ͠ఆظ࣮ߦ͢Δ • ……
• SQL Λී௨ʹୟ͘ • ෳͷσʔλϕʔεʹ·͕ͨͬͯूܭ͍ͨ͠? • SELECT COUNT(*) ʹ͔͔࣌ؒΔΑ͏ͳςʔϒϧͰͱʹ͔͘ू ܭ͍ͨ͠?
• ूܭΫΤϦΛߴԽ͢ΔͨΊʹՃͰ index ཉ͍͚͠Ͳ ͱΓ͋͑ͣݟͯΈ͍ͨϨϕϧͷूܭͷͨΊʹ ALTER TABLE Ͱ͖Δ? • ྺ࢙తܦҢͰγϦΞϥΠζ͞Εͨσʔλ͕ೖͬͯΔΧϥϜ͕͋Δ ͷΛΫΤϦ͍ͨ͠?
• SQL Λී௨ʹୟ͘ • ୟ͍ͨ݁Ռͷڞ༗ՄࢹԽͱ͔Ͳ͏͢Δ? • ॻࣺͯεΫϦϓτΛॻ͍ͯूܭͨ͠ͷΛεϓϨου γʔτʹΞοϓϩʔυ • ͦ͏͍͏ͷΛόονͱͯ͠ఆظ࣮ߦ͢Δ
• ͦͷεϓϨουγʔτ୭͕ϝϯςφϯε͢Δ? • …………
!!! BigQuery !!! • ͏·͍͍͍҆ۜͷؙ • సૹʹ Embulk Λ͍ɺ DSL
ͷதͰσʔλΛ͔ͯ͠Β όϧΫϩʔσΟϯά͢Δ • ඞཁͳσʔλશ෦ूΊ͖ͯͨΒવ JOIN ͳͲͰ͖Δ • σʔλྔ͕͍ͨͨ͜͠ͱͳ͍[ͲΕ͘Β͍?]ͳΒߟ͑ͳ͍ͱ͍ ͚ͳ͍͜ͱҰͳ͍
ΛϫϯγϣοτͰ͏ • Embulk ͰσʔλૹΔ • 1ճ͖ΓͳΒߟ͑ͳ͍ͱ͍͚ͳ͍͜ͱগͳ͍ • (ఆظ࣮ߦɾߋ৽ͷ͘͠ΈɺΠϯΫϦϝϯλϧϩʔυɺ……) • BigQuery
ͰͳΜΒ͔ूܭ͢Δ • Google εϓϨουγʔτʹΤΫεϙʔτ͢ΔϘλϯ • σʔλϙʔλϧ (Data Studio) ʹಥͬࠐΉ
BigQuery ʹΠϯϙʔτ͠ʹ͍͘ςʔϒϧ • ׂͷׂςʔϒϧΛ͍ʹ͍͘ςʔϒϧ • Ϩίʔυ͕ INSERT ͞Εͨ࣌ࠁɺΛද͢ΧϥϜ͕ແ͍ • ্Ґ֓೦ͷͦ͏͍͏ใΛ
JOIN ͖ͯͯͦ͠ΕͰύʔςΟγϣ ϯ͢Δ? • αʔϏεͷྺ࢙͕ 4000 (≒ 10.96 ) Ҏ্͋Δ • צͰύʔςΟγϣϯͷִؒΛؒҾ͘? • Α͘Α͘ߟ͑ͨΒຊ·ͩύʔςΟγϣχϯάඞཁͳ͍ͱ͔
Google BigQuery ྉۚ • ౦ژϦʔδϣϯͳΒετϨʔδ୯Ձ S3 ΑΓ͍҆ • 0.023USD /
GB / ݄ • ΫΤϦྉۚ 8.55 USD / 1 TB • ࠓͷҝସϨʔτͩͱ͍͍ͩͨ 1 GB ͋ͨΓ 1 ԁऑ • ͪΐͬͱ͍ͨ͠ํͳΒ΄ͱΜͲޡࠩϨϕϧ
͜͜·Ͱͷ·ͱΊ • αʔϏεͷσʔλʹରͯ͠ϔϏʔͳूܭΛͨ͘͠ ͳͬͨͱ͖ͷಀ͛ಓͱͯ͠ͷ Google BigQuery Λ ͝հ͠·ͨ͠
αʔϏεͷσʔλϕʔε͔Βूܭ ϩάϑΝΠϧ܈͔Βूܭ
ͦͷϩάࠓͲ͜ʹ • Ͳ͔͜ͷαʔό্? • ͦΕͦ͜ BigQuery ͱ͔ RedShift ͱ͔? •
Amazon S3 ʹஔ͔Ε͍ͯΔ? • ࣗͷ߹ݟ͔ͨͬͨͷ͕ S3 ʹஔ͔Εͯ ͍ͨ
S3 ʹஔ͍ͯ͋ΔͳΒ Amazon Athena Ͱ • S3 ʹஔ͍ͯ͋ΔϑΝΠϧΛΫΤϦͰ͖ΔϚωʔδυαʔϏε • த
Presto (https://prestodb.io/) • Presto Hive ͷ͕ࣝ͋Δͱ͍͜ͳͤΔͱ͍͏͜ͱ͕ଟ͍ • ϑϧϚωʔδυͳͷͰɺूܭͷͨΊʹετϨʔδͱωοτϫʔΫ ڧΊͷϚγϯΛ༻ҙͨ͠Γɺͱ͍ͬͨ͜ͱΛ͠ͳ͍͍ͯ͘ • ͓ΉΖʹ S3 ͷϑΝΠϧΛࢦఆͯ͠ CREATE TABLE ͯ͠ΫΤ Ϧ͢Δ͚ͩ
• ҙͷϑΥʔϚοτΛύʔεͯ͠ SQL ͰΫΤϦ • CSV/TSV, JSONL, ֤छΧϥϜφ (ྻࢤ) ϑΝΠϧͳͲ
ඪ४αϙʔτ • LTSV ਖ਼نදݱͰύʔεͯ͠ΫΤϦͰ͖Δ • → ਖ਼نදݱͰύʔεͯ͠Λൈ͖ग़ͤΔϑΝΠϧͳ ΒԿͰΫΤϦͰ͖Δ • ύʔε͠ͳ͍͜ͱͰ grep as a service ͱͯ͠׆༻
-- LTSV Λ Athena ͰΫΤϦ͢ΔͨΊʹύʔεͯ͠ΈΑ͏ CREATE EXTERNAL TABLE IF NOT
EXISTS `sugoi-service-logs` ( `time` string, `host` string, `method` string, `path` string, `taken_sec` float, `ua` string, `referer` string ) PARTITIONED BY (date string) ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.RegexSerDe' WITH SERDEPROPERTIES ( 'serialization.format' = '1', 'input.regex' = '^time:([^\\t]*)\\thost:([^\\t]*)\\tmethod:([^\\t]*)\\turi: ([^\\t]*)\\ttaken_sec:([^\\t]*)\\tua:([^\\t]*)\\treferer:([^\\t]*)$' ) LOCATION 's3://your-logs-bucket/sugoi-service-logs/' TBLPROPERTIES ('has_encrypted_data'='false')
-- LTSV Λ Athena ͰΫΤϦ͢ΔͨΊʹύʔεͯ͠ΈΑ͏ CREATE EXTERNAL TABLE IF NOT
EXISTS `sugoi-service-logs` ( `time` string, `host` string, `method` string, `path` string, `taken_sec` float, `ua` string, `referer` string ) PARTITIONED BY (date string) ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.RegexSerDe' WITH SERDEPROPERTIES ( 'serialization.format' = '1', 'input.regex' = '^time:([^\\t]*)\\thost:([^\\t]*)\\tmethod:([^\\t]*)\\turi: ([^\\t]*)\\ttaken_sec:([^\\t]*)\\tua:([^\\t]*)\\treferer:([^\\t]*)$' ) LOCATION 's3://your-logs-bucket/sugoi-service-logs/' TBLPROPERTIES ('has_encrypted_data'='false')
-- LTSV Λ Athena ͰΫΤϦ͢ΔͨΊʹύʔεͯ͠ΈΑ͏ CREATE EXTERNAL TABLE IF NOT
EXISTS `sugoi-service-logs` ( `time` string, `host` string, `method` string, `path` string, `taken_sec` float, `ua` string, `referer` string ) PARTITIONED BY (date string) ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.RegexSerDe' WITH SERDEPROPERTIES ( 'serialization.format' = '1', 'input.regex' = '^time:([^\\t]*)\\thost:([^\\t]*)\\tmethod:([^\\t]*)\\turi: ([^\\t]*)\\ttaken_sec:([^\\t]*)\\tua:([^\\t]*)\\treferer:([^\\t]*)$' ) LOCATION 's3://your-logs-bucket/sugoi-service-logs/' TBLPROPERTIES ('has_encrypted_data'='false') ……
ใ • S3 ʹͲΜͲΜϩάΛอଘ͢Δͱ͖ɺHive ܗࣜͷσΟϨΫτϦͰ อଘͯ͠ΔͱύʔςΟγϣχϯά͕ࣗಈͰߦΘΕͯศར • ྫ: s3://bucket/somelogs/year=2019/month=1/ day=23/engineer_seminar_001.log
• ͦ͏ͳͬͯͳͯ͘ symlink ͰͳΜͱ͔Ͱ͖Δͱ͍͏͋Δ • https://dev.classmethod.jp/cloud/aws/create-athena-table- with-symboliclink/
Amazon Athena ஈ • εΩϟϯͨ͠༰ྔʹରͯ͠՝ۚ (5 USD / 1 TB)
• gzip ѹॖࡁϑΝΠϧͦͷ··ΫΤϦͰ͖Δ • ྻࢤͰͳ͍ϑΝΠϧʹରͯ͠ΫΤϦ͢ΔͳΒɺूܭ ͢ΔΧϥϜΛߜͬͯྉۚઅʹͳΒͳ͍ • ج൫ͱͯ͠׆༻͍ͯ͘͠ײ͡ͳΒɺ ྻࢤϑΝΠϧʹม͢ΔϑΣʔζ͕ͳ͍ͱݫͦ͠͏
͜͜·Ͱͷ·ͱΊ • Amazon Athena Λ͏͜ͱͰɺ S3 ্ʹ͋ΔͳΒ͍͍ͩͨͲΜͳϩάͰͱ͘ʹ લ४උ͍͖ͤͣͳΓूܭͰ͖Δ
• όϧΫϩʔυͯ͠͠·͑ͬͪ͜ͷͷͱ͍͏ ؾ࣋ͪͰരूܭͰ͖Δάοζ • όϧΫϩʔυؾ͕ॏ͍ͱ͍͏ϩάϑΝΠϧͰɺ ϑΝΠϧͷ͋Γ͔Λઃఆ͢Δ͚ͩͰɺ Կ४උΛ͍͖ͤͣͳΓूܭͰ͖Δάοζ • ઌͷ݁ߏ͘͢ղܾͰ͖ΔΑ͏ʹ
• ઌͷΛղܾ͠ଓ͚ͯΔͱී௨ʹج൫ͬΆ͍ ͷ͕ཉ͘͠ͳͬͯ͘Δ͠ɺ͜͏ͳͬͯΔͱྑ͍ ͔ͳͱ͍͏ͷ͕ݟ͑ͯ͘Δ • ج൫͕͋Δͷʹӽͨ͜͠ͱͳ͍ • ͍Ζ͍Ζࢼ͢͜ͱͰσʔλͱͷ͖߹͍ํͷ Πϝʔδ͕ΒΜͩͷྑ͔ͬͨ