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
NoSQL概論
Search
UENISHI Kota
September 11, 2015
Technology
900
3
Share
NoSQL概論
High-level overview on NoSQL
UENISHI Kota
September 11, 2015
More Decks by UENISHI Kota
See All by UENISHI Kota
Storage Systems in Preferred Networks
kuenishi
0
89
Metadata Management in Distributed File Systems
kuenishi
2
560
Behind The Scenes: Cloud Native Storage System for AI
kuenishi
2
450
Apache Ozone behind Simulation and AI Industries
kuenishi
0
460
Distributed Deep Learning with Chainer and Hadoop
kuenishi
3
1.3k
A Few Ways to Accelerate Deep Learning
kuenishi
0
1.2k
Introducing Retz
kuenishi
5
1.2k
Introducing Retz and how to develop practical frameworks
kuenishi
3
810
Formalization and Proof of Distributed Systems (ja)
kuenishi
10
6.5k
Other Decks in Technology
See All in Technology
Gradle×GitHub_ActionsでCI時間を約50%短縮 ジョブ分割の設計と落とし穴 / Cutting CI Time by ~50% with Gradle and GitHub Actions: Job-Splitting Design and Pitfalls
takatty
0
600
AI駆動開発が変える、大規模開発の前提 ーHuman in the Loop から Human on the Loop へ / AIE2026
visional_engineering_and_design
2
1.5k
OCI Oracle AI Database Services新機能アップデート(2026/03-2026/05)
oracle4engineer
PRO
0
120
トークン数だけでは測れない — Claude Code 組織展開の効果検証から学んだこと
makikub
0
110
『家族アルバム みてね』における インシデント対応との向き合い方 / Approach incident response in Family Album
kohbis
2
290
AIプラットフォームを運用し続けるための可観測性
tanimuyk
4
1k
ルールやカスタム機能、どう使う?理想の出力を引き出すために今知りたいIBM Bob 5つの機能
muehara
1
280
Sony_KMP_Journey_KotlinConf2026
sony
2
200
AI Engineering Summit Tokyo 2026 AIの前に、やることがある 〜医療データ企業の4フェーズ〜
dtaniwaki
0
970
ITエンジニアを取り巻く環境とキャリアパス / A career path for Japanese IT engineers
takatama
4
1.8k
AIガバナンス実践 - 生成AIコネクタのデータ漏洩リスクと実務対策
knishioka
0
160
Claude code Orchestra
ozakiomumkj
3
900
Featured
See All Featured
The Pragmatic Product Professional
lauravandoore
37
7.3k
Taking LLMs out of the black box: A practical guide to human-in-the-loop distillation
inesmontani
PRO
3
2.2k
The MySQL Ecosystem @ GitHub 2015
samlambert
251
13k
The Impact of AI in SEO - AI Overviews June 2024 Edition
aleyda
5
1.1k
個人開発の失敗を避けるイケてる考え方 / tips for indie hackers
panda_program
122
22k
Fight the Zombie Pattern Library - RWD Summit 2016
marcelosomers
234
17k
Bridging the Design Gap: How Collaborative Modelling removes blockers to flow between stakeholders and teams @FastFlow conf
baasie
0
570
Hiding What from Whom? A Critical Review of the History of Programming languages for Music
tomoyanonymous
2
840
Designing for Timeless Needs
cassininazir
1
240
A designer walks into a library…
pauljervisheath
211
24k
Reflections from 52 weeks, 52 projects
jeffersonlam
356
21k
Refactoring Trust on Your Teams (GOTO; Chicago 2020)
rmw
35
3.5k
Transcript
NoSQL ֓ 2015/9/11 DB Tech Showcase Sapporo Basho δϟύϯ᷂ɹ্߁ଠ
୭ʁ • @kuenishi • Github, Twitter, etc • ࢄγεςϜྺ7 •
Bashoδϟύϯͷํ͔Βདྷ·ͨ͠ • Riak CSͷ։ൃ • ࠷ۙτϥϯβΫγϣϯʹڵຯ ͕͋Γ·͢
Ξϯέʔτ •SQL͡Όͳ͍ͱϦϨʔγϣφϧͳσʔλϕʔ ε࡞Εͳ͍ʁ •ʮཧ͔ΒֶͿσʔλϕʔε࣮ફೖʯΛಡ Μͩʁ •ʮNoSQLͷجૅࣝʯΛಡΜͩʁ •Dynamo? BigTable? Tandem?
ΞδΣϯμ •NoSQLΑ͏Θ͔ΒΜʂԿʁʂͱ͍͏ਓͷͨΊʹɺ ԿͳͷͬͯΛ͠·͢ •͍Ζ͍Ζ͋ͬͯΘ͔ΒΜʂͱ͍͏ਓͷͨΊʹྨ ࣠ʹ͍͓ͭͯ͠·͢ •ͦͷଞɺ༩ଠΛ͠·͢
Relational DBMS •Ұൠతʹ… •ෳͷΫϥΠΞϯτ͔ΒTCP/IPͰଓ͠ •ϦϨʔγϣφϧϞσϧΛ࣮༻Խͨ͠SQLͱ͍͏ݴޠΛΠϯλʔ ϑΣʔεͱ͠ •B+treeͱϒϩοΫΩϟογϡͱ͍͏σʔλߏΛ͍ɺ •WALͱB+treeͰӬଓԽΛ͠ʢϨϓϦέʔγϣϯ͠ͳ͍ʣɺ •MVCCͳͲͰฒߦੑ੍ޚΛߦ͏σʔλϕʔεཧγεςϜ
NoSQL? •Not Only? •ݴ༿ͱҟͳΔɺ2௨Γͷ࣮ଶ •A) σʔλϞσϧɺΠϯλʔϑΣʔεͱͯ͠SQLΛͬͯ ͍ͳ͍ •B) طଘͷτϥϯβΫγϣϯॲཧٕज़ͰͰ͖ͳ͔ͬͨ͜ ͱΛ࣮ݱ͢ΔͨΊʹɺผٕज़Λ࠾༻ͨ͠
SQLͰͳ͍ʁ •ϦϨʔγϣφϧϞσϧͰͳ͍ԿΒ͔ͷσʔλϞσϧ •σʔλදݱΫΤϦݴޠʹSQLΛΘͳ͍ •υΩϡϝϯτDB: “Semi-structured database” •Redis •KVS (Key-Value Store):
ฆΒΘ͍͠
Document DB •ͱͱͷࢥ: •͋ΒΏΔλΠϓͷυΩϡϝϯτΛอଘͯ͠ΫΤϦ Ͱ͖ΔΑ͏ʹ͠Α͏ •JSON, XML, PDF, etc… •ݱ࣮:
݁ہ… •KVS ʹ JSON + ΠϯσοΫε + ࢄMapReduce
ACIDͰͳ͍ʁ
ACID •Atomicity •ߋ৽ͷ్தঢ়ଶ͕ݟ͑ͳ͍͜ͱ •Isolation •ଞͷਓ͕ߋ৽͍ͯ͠Δͷ͕ݟ͑ͳ͍͜ͱ •Key-Value StoreͳΒɺ୯ҰͷΤϯτϦΛϩοΫͯ͠ߋ৽͢Δͩ ͚ͳͷͰAtomicͩ͠Isolated!! ʢϗϯτʁʣ
ACID •RDBͩͱ •ϦϨʔγϣϯಉ࢜ͷσʔλ͕߹͍ͯ͠Δ͜ͱ •ࢄτϥϯβΫγϣϯͯ͠σʔλ͕ෆ߹͠ͳ͍͜ͱ •ނোͯ͠߹ͨ͠ঢ়ଶͰϦΧόϦͰ͖Δ͜ͱ •NoSQLͩͱ •ϨϓϦέʔγϣϯͯ͠σʔλ͕ෆ߹͠ͳ͍͜ͱ •Key-Value StoreͳΒ…
ACID •RDBͩͱ •ӬଓԽσΟεΫʹॻ͍ͨΒऴΘΓ •σΟεΫյΕͨΒΒͶ •NoSQLͩͱ •σΟεΫյΕͯιϑτΣΞͰԿͱ͔͢Δ •ࣗͰϨϓϦέʔγϣϯΔʂ
εέʔϧΞτʁ •σʔλͷۭؒΛԿΒ͔ͷϧʔϧʹଇׂͬͯ͢Δ •ׂͨ͠σʔλΛࢄͯ͠ཧ͢Δ •ϊʔυΛ૿͢ͱॲཧೳྗ༰ྔ͕૿͑Δʢʁʣ
εέʔϧΞτਤղ σʔλۭؒ ׂۭؒ
B+Tree •ΩʔΠϯσοΫε͔Β O(log N) ҎԼͷίετͰ σʔλΛݟ͚ͭΔͨΊͷσʔλߏ •εέʔϧΞτ͢Δͱ… •͍͔ͭ͘ิʢସʣٕज़͕͋Δ •Consistent Hashing
•γϟʔσΟϯά (Range Partitioning) •֊γϟʔσΟϯά
εέʔϧΞτͱނোੑ •Q. 1ͷϊʔυͷނো֬Λ p ͱ͢Δͱɺ n ͷΫϥελ͕શͯ ਖ਼ৗಈ࡞͍ͯ͠Δ֬ʁ 1. 1
- p^n 2. p^n 3. (1 - p)^n •εέʔϧ͕େ͖͍ͱɺͲ͔͕͜΄΅ඞͣյΕ͍ͯΔঢ়ଶʹͳΔ •෦ނো͕ى͖͍ͯͯશମ͕ਖ਼͘͠ಈ࡞͢Δઃܭ͕ඞཁ •ϨϓϦέʔγϣϯʹΑΔࣗಈతͳϑΣΠϧΦʔόʔͳͲ͕ඞཁ
ਖ਼ৗಈ࡞ͷ֬ n \ p 0.1 0.01 0.001 1 0.9 0.99
0.999 10 0.348678440 0.904382075 0.990044880 100 0.000026561 0.366032341 0.904792147 1000 0 0.000043171 0.367695424 ※༗ޮࣈ…
εέʔϧΞτਤղ 2 σʔλۭؒ ෳ
ӬଓԽͱ •ܗ͋Δͷ͍͔ͭյΕΔ •ॾߦແৗʢେൠᔷᒒܦʣ •2ͭҎ্ͷίϐʔΛ͓࣋ͬͯ͘Ҏ֎ʹɺ σʔλΛӬٱʹอଘ͢Δํ๏ͳ͍ •ෳͷͳ͍σʔλϕʔεͳ͍
ෳඞͣ࡞ΒΕΔ •LANͷ͜͏Ͱ •SANͷ͜͏Ͱ •ւͷ͜͏Ͱ •ނো୯Ґ͕ඞ͔ͣΕ ͍ͯΔ (C) Evan Amos
ނোͯ͠ಈ࡞ܧଓ͢Δෳ •Ͳ͏͍͏ނোϞσϧΛલఏʹͯ͠Ϩϓ ϦέʔγϣϯΛΜ͔͕ͩॏཁ •Fail-stopΛલఏʹ͍͚ͯ͠ͳ͍ •ʮյΕ͔͚ͷRadioʯ •͍ΘΏΔαΠϨϯτނো •ωοτϫʔΫͷஅɺॠஅ •Disk Hiccup (C)
Evan Amos
ωοτϫʔΫͷஅ •அͨ͠ΒɺͲͪΒ͕ਖ਼͠ ͍͔ܾΊͳ͚Ε͍͚ͳ͍ •ϚχϡΞϧͰհೖ͢Εܾ ΊΒΕΔʢ͔͠Εͳ͍ʣ •ࣗಈԽ͢Δʹɺωοτϫʔ Ϋ͕ΕͯσʔλΛ߹ ͤ͞Δෳٕज़͕ඞཁ ? ?
(C) Evan Amos
ෳͷٕज़ ෳͷλΠϓ ނোͳͲͷલఏ Master-slave ಉظ ωοτϫʔΫ͕ނোͨ͠Βࢭ·Δ Master-slave ඇಉظ ωοτϫʔΫ͕ނোͨ͠Β σʔλ͕ෆ߹͢Δ͔
Consensus (Strong Consistency) ωοτϫʔΫ͕ނোͨ͠ΒҰ෦ಈ࡞ Eventual Consistency ωοτϫʔΫ͕ނোͯ͠શಈ࡞
ෳͰͳ͍σʔλͷ߹ੑ •εέʔϧΞτ͢Δͱɺผϊʔυ্ͷ ผͷσʔλΛ߹ͨ͠ঢ়ଶͰߋ৽͠ͳ ͚ΕͳΒͳ͍ •ࢄͨ͠ΠϯσοΫεͷߋ৽ •֎෦Ωʔͷߋ৽ •τϦΨʔͷॲཧ
εέʔϧΞτਤղ 3 ςʔϒϧ ΠϯσοΫε or FKey or else
ෳͰͳ͍σʔλͷ߹ੑ อো •2PC (or ࢄϩοΫ) •ϩοΫΛ࣋ͬͨ··ϊʔυ͕ނোͨ͠ͱ͖ͷѻ͍ •Cluster-wideͰҰ؏ͯ͠୯ௐ૿Ճ͢ΔԿΒ͔ͷΛੜ͢Ε ɺTSOMVCCΛ࣮Ͱ͖Δ •λΠϜελϯϓ …
࣌ܭͷಉظΛอো͢Δํ๏͕ࣗ໌Ͱͳ͍ •IDΛൃߦ͢ΔϚελʔϊʔυ … SPoF •͜͜·ͰͪΌΜͱ͍ͬͯΔNoSQL΄ͱΜͲͳ͍
NoSQLͷऴΘΓ •RDBMSͰղ͚ͳ͔͕͍͔ͬͨͭ͘ղ͚ͨͷͰʮͬͺΓSQL͕΄͍͠ʯ ʮ׳Εٕͨज़ऀ͕ଟ͍ͷͰʯ •ϦϨʔγϣφϧϞσϧʹଇ͍ͯ͠ͳ͍σʔλͰSQLͰΫΤϦΛॻ͚ͯ͠·͏ •͍͔ͭ͘ͷNoSQLɺτϥϯβΫγϣϯཧΛͦͦ͜͜ʹ SQL(-like ͳ) ΠϯλʔϑΣʔεΛఏڙͭͭ͋͠Δ •CQL, N1QL,
etc.. •OLAP͚ͷMPPΤϯδϯSQL͕ඪ४ •Impala, Drill, PrestoDB, Calcite
2ͭͷٕज़ •ΫΤϦॲཧ •RDBMSͰSQL •RDBMSͰͳͯ͘SQL •τϥϯβΫγϣϯॲཧ •ACIDΛ࣮ݱ͍ͯ͠ΔRDBMS •ࢄ͍ͯͯ͠ACIDΛࢦ͍ͯ͠Δ
Summary •ʢੲʣNoSQL ͱɺطଘͷσʔλϕʔεٕज़Λ1͔Βݟͳ ͓ͯ͠ɺRDBMSͰͰ͖ͳ͔ͬͨ͜ͱΛୡ͠Α͏ͱͨ͠Ұ ࿈ͷࢼΈ •ϦϨʔγϣφϧϞσϧɺB+treeɺϨϓϦέʔγϣϯ •ʢࠓʣ͏Կ͕ͳΜ͔͔ͩΒͳ͍͘Β͍ଟ༷Խ •ٕज़తͳ؍ʢͷྨʣ •εέʔϧΞτɺނোੑɺ߹ੑʢɺ͋ΔΑ