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
アルとAlgoliaと私 / alu_algolia
Search
Sponsored
·
Ship Features Fearlessly
Turn features on and off without deploys. Used by thousands of Ruby developers.
→
Hikaru Tooyama
May 14, 2019
Business
2.5k
8
Share
アルとAlgoliaと私 / alu_algolia
2019-05-14 Algolia Community Party in Tokyo
Hikaru Tooyama
May 14, 2019
More Decks by Hikaru Tooyama
See All by Hikaru Tooyama
Firestore東京リージョン利用に伴う Firebaseプロジェクト移行手順 / alu-firestore #アル
vexus2
8
2.6k
Firestoreを本番運用して得た知見と事件簿 / manga-village
vexus2
6
2.6k
エンジニア向けサービスを提供するチームでの開発環境改善
vexus2
1
1.3k
Other Decks in Business
See All in Business
suisei.inc_ company deck
suisei2015
0
370
RecruitingGuide(KR)
kakaojapan
0
650
紹介パートナー様向け 紹介手数料プランとご登録手順のご案内(マルコポーロ)
kimete
0
160
BizMow会社紹介資料_2026
bizmow
0
300
エージェントスキルによる最適化
mickey_kubo
2
140
哲学ドリブン開発の全体像 ── 同じプロンプトで出力が変わるとき、何が起きているのか~Philosophy as Code
makitotashiro
0
170
01_全社_FLUX採用ピッチ資料_Ver.5.3
flux
PRO
8
210k
Claude Codeで毎日のToDoとShould to doを配信させる方法
zashii
0
160
Clarity for Product People
arnekittler
0
350
“使われているハーネス/使われていないハーネス”を可視化するところから始めた話
sugamoto
0
190
YassLab (株) サービス紹介 / Introduction of YassLab
yasslab
PRO
3
43k
インターセクト会社説明資料
intersect
0
180
Featured
See All Featured
The Organizational Zoo: Understanding Human Behavior Agility Through Metaphoric Constructive Conversations (based on the works of Arthur Shelley, Ph.D)
kimpetersen
PRO
0
340
Documentation Writing (for coders)
carmenintech
77
5.4k
CSS Pre-Processors: Stylus, Less & Sass
bermonpainter
360
30k
Claude Code どこまでも/ Claude Code Everywhere
nwiizo
65
55k
What’s in a name? Adding method to the madness
productmarketing
PRO
24
4.1k
How to train your dragon (web standard)
notwaldorf
97
6.6k
I Don’t Have Time: Getting Over the Fear to Launch Your Podcast
jcasabona
34
2.8k
Lightning Talk: Beautiful Slides for Beginners
inesmontani
PRO
2
560
Chasing Engaging Ingredients in Design
codingconduct
0
200
The AI Search Optimization Roadmap by Aleyda Solis
aleyda
1
5.8k
B2B Lead Gen: Tactics, Traps & Triumph
marketingsoph
0
130
Reflections from 52 weeks, 52 projects
jeffersonlam
356
21k
Transcript
1 ΞϧͱAlgoliaͱࢲ Ξϧגࣜձࣾɹԕࢁߊ !WFYVT
2 ࣗݾհ - Ξϧגࣜձࣾ ԕࢁߊ (@vexus2) - ݱ৬Ͱ VP of
Product - גࣜձࣾStorm දऔక - ։ൃίϯαϧςΟϯάۀ(Firebaseͱ͔) - ݩʑBackend(Ruby/PHP)͕ͩ࠷ۙFrontendدΓ - TypeScript / Vue / Nuxt / Firebase ͳͲ - Algoliaྺ16ϲ݄͘Β͍ - ݸਓαʔϏε࡞Δͱ͖ຖճAlgoliaʹσʔλΛೖΕΔͱ͜Ζ͔Β࢝ΊΔ ͘Β͍ͷAlgoliaґଘϢʔβ - झຯϚϯΨΛಡΉ͜ͱ - ࠓಡΜͰ໘ന͔ͬͨͷʮਖ਼ෆಈ࢈ʯʮఱࠃେຐڥʯʮΞΫλʔδϡʯʮONE OUTSʯʮϥϯ ΣΠͰসͬͯʯʮͦͷணͤସ͑ਓܗ࿀Λ͢ΔʯʮαπϦΫϧʔτʯʮͬ͠ΆͷʯʮΑฉ͍ͯ͘Εʯ ʮಓͷѱঁͨͪʯʮβɾϑΝϒϧʯʮ339ʯʮϰΟϯϥϯυɾαΨʯ
3 ࣗݾհ ແྉͰఏڙ͞Ε͍ͯΔອըΛूΊͨʮອըϏϨοδʯΛݸਓͰӡӦɻ ϐʔΫ࣌ͷաڈ30ؒͰ1060ສPV͕͋ΓɺͦͷޙΞϧϦϒϥϯυɻ Xxx ίϛϡχςΟ GoogleͰʮອըଜʯͰݕࡧ͢Δͱ1Ґʹ Yahoo!χϡʔεͷτοϓʹ(هࣄݩɿITmedia)
4 Ξϧͱʁ W
5 Ξϧʹ͍ͭͯ ΞϧϚϯΨϑΝϯͷͨΊʹɺϚϯΨΛൃݟͰ͖ΔศརͳαʔϏεΛఏڙ͢Δ ϚϯΨ୳͠ͷใαʔϏεͰ͢ɻ iOS/Web൛ΛͦΕͧΕఏڙதɻ(Android͏͙͢!)
6 Ұ෦ͷϚϯΨͷίϚ͕ߘɾӾཡ͕Ͱ͖·͢ɻ ֤ग़൛ࣾʹʹڐΛΒ͍ͬͯΔͷͰɺ ߹๏తͳίϚͷߘɾӾཡ͕Ͱ͖ΔαʔϏεͰ͢ɻ Ξϧʹ͍ͭͯ
7 શੈքதʹʮϚϯΨͷσʔλϕʔεʯͱ͍͏ͷ͕ଘࡏ͠ͳ͍ͷͰɺͱʹ͔͘େྔ ͳϚϯΨσʔλΛటष͘ूΊ͍ͯΔσʔλϕʔεͱͯ͠ͷଆ໘͋Γ·͢ɻ ʢΫϩʔϥʔɾೖߘɾਓྗͳͲʣ Ξϧʹ͍ͭͯ
8
9 FirestoreͱAlgoliaͷڞଘ W
10 Firestoreͷલఏ 1. FirestoreFirebaseͷυΩϡϝϯτࢦܗͷεέʔϥϒϧͳNoSQL 2. Queryׂ͕ͱශऑ - `<` >` `==`
`>=` `<=` ͑Δ - != (not equal) ͕ ͑ͳ͍ 3. limit, offsetͷΑ͏ͳ͜ͱ͕Ͱ͖ͳ͍(startAfter, startAt)ͷͰɺ ϖʔδωʔγϣϯͰͷpage=2ͷΑ͏ͳURLͷߏங͕ࠔ
11 Firestoreͷલఏ 4. औಘܶతʹ͘ͳ͍ɻ ࣮ߦڥ ॳճ 2ճҎ߱ us-central 1݅औಘ 892.3ms
202.3ms 20݅औಘ(SubCollection) 864ms 260.8ms asia-northeast1 1݅औಘ 370.3ms 125.6ms 20݅औಘ(SubCollection) 461.3ms 156.4ms ࢼߦճ100ճఔͷ؆қతͳϕϯνϚʔΫͰͷฏۉɻ ࣮ߦωοτϫʔΫڥͳͲʹΑͬͯมΘΔ͕ࢀߟ·Ͱɻ
12 Firestoreͱͷ͍͚
13 Firestore or Algolia? - ͋͘·Ͱσʔλϕʔεͱͯ͠Firestore͕Ϛελ - UniqueͳIDΛࢦఆͯ͠Ұҙʹऔͬͯ͜ΕΔ߹ʹFirestore - ୯ҰϑΟʔϧυͰͷSortQueryͰऔͬͯ͜ΕΔ߹ʹFirestore
- ෳࡶͳQuery݅ಛघͳSort݅(offsetͱؚ͔Ή)ɺPaginationΛؚΉ ߹ʹAlgolia - (Web൛ͳͲͰ)σʔλऔಘͷΛٻΊΔͱ͖ʹAlgolia
14 Ͳ͏͍͏σʔλΛAlgoliaʹ ͚ͬΔ͔
15 σʔλొʹ͍ͭͯ - AlgoliaݕࡧͱผͰϨίʔυͰͷ՝ۚ͋ΔͷͰʮݕࡧʹඞཁͳ࠷ খݶͷςʔϒϧʯͷΈొ - λΠϛϯάCloud Functions(Firebase)ͰFirestoreͷߋ৽τϦΨʔͷؔ Ͱɺߋ৽͞ΕͨσʔλΛAlgoliaʹొ͍ͯ͠Δ -
σʔλෆ߹͕ΠϠͳͷͰɺجຊతʹFirestoreͱಉ͡σʔλΛAlgolia ଆʹίϐʔ͢Δ
16 Algoliaͷخ͍͠ͱ͜Ζ W
17 خ͍͠ͱ͜Ζ: ͱʹ͔͍͘ - ͱʹ͔͘Response͕͍ɻΘ͚͕Θ͔Βͳ͍͘Β͍ʹɻ - 1୯ҐͷMonitoring reportݟͯɺaverage 1ms /
99th percentile 3ms ͱ͔͕βϥʹ͋Δ - աڈ1ͰҰ൪͔ͬͨͷ͕ average5ms / 99th percentile 16ms
18 خ͍͠ͱ͜Ζ: ݕࡧΦϓγϣϯ͕๛ - શҰகɺશจҰகͳͲͷݕࡧΦϓγϣϯ͕๛ͳͷͰʮHITݮΒͯ͠ Ͱਫ਼ͷߴ͍ݕࡧαδΣετʯͷΑ͏ͳ͜ͱ͕ϩδοΫଆͰௐ͢͠ ͍ - `matchLevel` ͱ͍͏ύϥϝʔλ͕ฦ͞ΕΔͷͰશҰகͷ߹͍Λஅ
Ͱ͖Δ - ʮDEATH NOTE (େ͙ͭΈɾখാ݈)ʯͷஶऀؔ࿈࡞ͱͯ͠ҎԼͷΑ͏ ͳ੍ޚָ͕ʹͰ͖Δ - ʮόΫϚϯ(େ͙ͭΈɾখാ݈)ʯදࣔ͢Δ - ʮϓϥνφΤϯυ(େ͙ͭΈɾখാ݈)ʯදࣔ͢Δ - ʮώΧϧͷޟ(΄ͬͨΏΈɾখാ݈)ʯදࣔ͠ͳ͍ - ʢදࣔ͢Δɺͱ͍͏Φϓγϣϯ༰қʹͰ͖Δʣ
19 خ͍͠ͱ͜Ζ: Web্Ͱ΄ͱΜͲΧελϚΠζͰ͖Δ - ݕࡧରͷΧϥϜ໊ɺrank͚ɺσϑΥϧτιʔτɺtypoܥͷڐ༰ൣғɺ Synonym(ಉٛޠ)ɺ۠Γจࣈઃఆɺϖʔδωʔγϣϯઃఆɺresponse ϑΟʔϧυઃఆɺetc… ͕Web্͔ΒઃఆͰ͖Δ
20 خ͍͠ͱ͜Ζ: ͍͍ײ͡ͷϨϙʔτ͕σϑΥϧτͰऔΕΔ - ຖͷݕࡧ No Result Rate, ্Ґݕࡧ No
Result Search ͷ݁ՌͳͲ ͕Index୯ҐͰࣗಈతʹAnalyticsʹೖ͍ͬͯΔɻ - Ξϧͷ߹ʮݕࡧͰHIT͠ͳ͔ͬͨϚϯΨʯͷ্ҐΫΤϦΛݩʹσʔλͷ ֦ॆͳͲͷ؍Ͱ͓͏ͱ͍ͯ͠Δ
21 AlgoliaͷͭΒ͍ͱ͜Ζ W
22 ͭΒ͍ͱ͜Ζ: ຊޠදهΏΕ͕བྷΉͷݸผରԠ - ຊޠͷΏΒ͗ͦ͜·ͰNo CustomizeͰڧ͘ͳ͍ͷͰɺ SynonymΛઃఆ͢Δඞཁ͕͋Δ - ONE PIECEΛʮϫϯϐʔεʯʮΘΜͽʔ͢ʯʮƂƃűƅʯʮONEPIECEʯ
ʮϫϯͽʯΈ͍ͨʹදهΏΕ͕େ͖͍ͱσϑΥϧτͷঢ়ଶͰతͷ ͷ͕HIT͠ͳ͍ - ΞϧͰSynonymΘͣʹϚϯΨͷʮུশʯʮදهΏΕʯٵऩ༻ͷΧ ϥϜ(alias)Λผ్ఆٛͯ͠Algoliaଆʹಉظ͍ͤͯ͞Δɻ
23 ͭΒ͍ͱ͜Ζ: filterपΓ͕ෳࡶ - ORݕࡧANDݕࡧͳͲ݅Λࢦఆ͢ΔʹfilterΛ͏͜ͱʹͳΔ͕ɺएׯ Ϋη͕͋ͬͯ࠷ॳ͔ΓͮΒ͍ɻ - ݅(filter)Λࢦఆ͢Δ߹ʹࣄલʹ Facet(attributesForFaceting)Λઃఆ ͢Δඞཁ͕͋Δ
- AND OR filtersͷதͰϕλͰॻ͘ඞཁ͕͋ΔɻएׯՄಡੑѱ͍ɻ - Boolean ܕ ΛFilter݅ʹೖΕΔͱ͖ʹ true/false Ͱͳ͘ 1/0 Ͱ͢ ඞཁ͕͋Δ
24 ͭΒ͍ͱ͜Ζ: Φϓγϣϯܥ͕߲͕ଟ͍ - QueryपΓΦϓγϣϯ͕๛ʹ͋ΔͷͰɺͰ͖Δ͜ͱΛશ෦Ѳ͢Δͷ͕ େมɾɾɾ - prefixType? advancedSyntax? optionalWords?
ͳʹͦΕʁঢ়ଶʹͳΔ - ෳͷΦϓγϣϯઃఆͨ͠ͱ͖ʹͲ͏͍͏݁ՌʹͳΔͷ͔ʁ͕͛Δલ ʹ૾͠ʹ͍͘͜ͱ͕ଟ͍ɻͷͰɺͻͨ͢Βࢼ͢͜ͱ͕ଟ͍ɻ - ͱʹ͔͘ࡶʹσʔλΛಥͬࠐΜͰऔಘ͢Δ͚ͩͳΒ؆୯ɻෳࡶͳ͜ͱ Ζ͏ͱ͢ΔͱυΩϡϝϯτΛख़ಡ͢ΕେମͳΜͱ͔ͳΔɻ
25 ͭΒ͍ͱ͜Ζ: Ϣʔβ͝ͱͷηΩϡΞͳσʔλ - جຊతʹશϨίʔυ͕ݕࡧରʹͳΔͷͰɺྫ͑ʮϢʔβ͝ͱͷߪೖ ཤྺʯͳͲଞऀͷݕࡧ݁Ռʹදࣔͨ͘͠ͳ͍ͷѻ͍ʹ͍͘ - Solutionͱͯ͠ʮࣗͷߪೖͨ͠ͷ͔͠ݟ͑ͳ͍(=ৗʹࣗͷuserIdͰ ϑΟϧλ)ʯݕࡧʹͳΔAlgoliaͷAPI_KEYΛϢʔβ͝ͱʹൃߦ͢ΔॲཧΛ༻ ҙͯ͠ɺϢʔβ͝ͱʹ֤ࣗͷΩʔͰݕࡧ͢ΔΑ͏ʹ͢ΕՄೳ
- ॲཧ͕͔ͳΓࡶʹͳΔ͠API_KEYͷऔΓճ͠Ͱ࣌ؒऔΒΕΔͷͰ Algoliaͷ͕٘ਜ਼ʹͳΔͷͰɺ͋·Γϕλʔͳײ͡͠ͳ͍ https://firebase.google.com/docs/firestore/solutions/search?hl=ja
26 ·ͱΊ W
27 ·ͱΊ - ͍ɾՁ֨ʢ͔Β͑Δʣɾߴػೳͱࡾഥࢠἧ͍ͬͯΔɻݸਓنͰ શવ͑Δஈଳʢͱ͍͏͔Freeϓϥϯ͋ΔʣͳͷͰϢʔβɾγεςϜ Θͣݕࡧͱ͍͏ߦҝΛߦ͏αʔϏεͰ͋ΕΘͳ͍ཧ༝ͳ͍ɻ - Algolia͍͔Βͦ͜ґଘੑ͕ߴ͍αʔϏεͳͷͰɺͲΜͲΜϨίʔυ ೖΕͯΨϯΨϯݕࡧ͢Δͱ͓ஈͦͦ͜͜ߴ͘ͳΓ͕ͪɻ͝ར༻ܭը తʹ
28