Realtime Database for high traffic production application

Slide 1

Slide 1 text

Realtime Database for high traﬃc production application @sota1235 2018/9/1 GDG DevFest Tokyo 2018

Slide 2

Slide 2 text

No content

Slide 3

Slide 3 text

No content

Slide 4

Slide 4 text

For what?

Slide 5

Slide 5 text

No content

Slide 6

Slide 6 text

Mercari channel • Live e-commerce • Working on Mercari app • You can buy/sell anything through streaming https://www.mercari.com/jp/mercari-channel/

Slide 7

Slide 7 text

• Messages • Likes • Notiﬁcations • Item list

Slide 8

Slide 8 text

Today • Why Realtime Database? • How to build app • How to run app on production

Slide 9

Slide 9 text

Note • Today, Iʼll share some techniques • I believe itʼll help you to use Realtime Database

Slide 10

Slide 10 text

Why Realtime Database?

Slide 11

Slide 11 text

Tight schedule Easy to use Sync data in real time What we needed

Slide 12

Slide 12 text

Points • Fully managed storage • Powerful SDK • Sync data in real time

Slide 13

Slide 13 text

Fully managed • 1 JSON data in 1 database • Multiple databases per 1 Firebase Project • Make project and ship it

Slide 14

Slide 14 text

Powerful SDK https://ﬁrebase.google.com/docs/guides/

Slide 15

Slide 15 text

// Set the configuration for your app const config = { apiKey: "apiKey", authDomain: "projectId.firebaseapp.com", databaseURL: "https://databaseName.firebaseio.com", storageBucket: "bucket.appspot.com" }; firebase.initializeApp(config); // Get a reference to the database service const database = firebase.database(); Initializing

Slide 16

Slide 16 text

const starCountRef = firebase.database().ref('posts/starCount'); starCountRef.on('value', (snapshot) => { console.log(snapshot.val()); });

Slide 17

Slide 17 text

Sync data in real time Subscribe Publish

Slide 18

Slide 18 text

How to build app

Slide 19

Slide 19 text

Architecture design Schema design

Slide 20

Slide 20 text

Architecture design Architecture design Schema design

Slide 21

Slide 21 text

Read Write ・Using SDK to read data ・Techniques not to read all data ・Updating data with REST API ・Managing frequency of query

Slide 22

Slide 22 text

Read data • Getting data through SDK • Thatʼs it • We need only 1 hack

Slide 23

Slide 23 text

$PNNFOU $PNNFOU $PNNFOU Time

Slide 24

Slide 24 text

$PNNFOU $PNNFOU $PNNFOU Time Start seeing streaming

Slide 25

Slide 25 text

$PNNFOU $PNNFOU $PNNFOU Time Start seeing streaming SDK got this even though it’s unnecessary ☹

Slide 26

Slide 26 text

Little hack • SDK gets one data when connecting • But sometimes itʼs unnecessary • So ﬁltering data by timestamp

Slide 27

Slide 27 text

Write data • Updating data through REST API • Managing frequency of query

Slide 28

Slide 28 text

No content

Slide 29

Slide 29 text

No content

Slide 30

Slide 30 text

REST API • Two ways to write data • REST API or WebSocket(w/SDK) • In our case, REST API matched

Slide 31

Slide 31 text

Authentication • Database token • Deprecated • Google OAuth2 • Recommended

Slide 32

Slide 32 text

Cache auth token • Cache auth token on cache store • In case of sending request to REST API, it will work • OAuth2 authentication will be overhead

Slide 33

Slide 33 text

keep-alive with chocon • We use chocon for using keep-alive • PHP process canʼt use it • Realtime Database instances are in US • 1 request means RTT to the Paciﬁc Ocean • https://github.com/kazeburo/chocon

Slide 34

Slide 34 text

Keep alive with chocon + , keep-alive

Slide 35

Slide 35 text

Frequency managing • Not sending all data from user

Slide 36

Slide 36 text

Frequency managing • We donʼt need to send these kinds of data every time • Likes, Audience count, Notiﬁcations • So we manage frequency of sending such data

Slide 37

Slide 37 text

Mercari API Like count=1 count=1

Slide 38

Slide 38 text

Mercari API Like count=1 count=1 Like Like Like count=2 count=3 count=4

Slide 39

Slide 39 text

Mercari API Like count=1 count=1 Like Like Like count=2 count=3 count=4

Slide 40

Slide 40 text

Mercari API Like count=1 count=1 Like Like Like count=2 count=3 count=4 Check frequency before sending data N query/sec

Slide 41

Slide 41 text

Mercari API Like count=1 count=1 Like Like Like count=2 count=3 count=4 count=4 N sec

Slide 42

Slide 42 text

Motivation • We donʼt need to update UI 1000 times • The most important point is experience • Actually, spec is limited for 1 instance • 1000 update per sec

Slide 43

Slide 43 text

Architecture design Schema design Schema design

Slide 44

Slide 44 text

Realtime Database is schema-less Then, how to design?

Slide 45

Slide 45 text

Rule • Settings for 1 database • Written in JSON • Validation • Permission/Authorization

Slide 46

Slide 46 text

{ "rules": { ".write": false, "lives": { "$live_id": { "messages": { "$message_id": { ".validate": "newData.hasChildren(['user', 'text'])" } }, "notifications": { "buy_item": { ".validate": "newData.hasChildren(['text'])" } } } } } } Sample rule

Slide 47

Slide 47 text

Our strategy • Only writing data from our server • Controlling read permission by using ﬂag data on database

Slide 48

Slide 48 text

Only writing from server • Pros • Not need to be careful about permission for client • Cons • Server needs to handle high traﬃc

Slide 49

Slide 49 text

Slide 50

Slide 50 text

Slide 51

Slide 51 text

Slide 52

Slide 52 text

Controlling permission dynamically • Rule can refer data on schema • It means we can manage permission by data

Slide 53

Slide 53 text

{ "rules": { ".write": false, "lives": { "$live_id": { "messages": { "$message_id": { ".validate": "newData.hasChildren(['user', 'image', 'text'])" } }, "notifications": { "buy_item": { ".validate": "newData.hasChildren(['text'])" } } } }, "alive_lives": { ".write": false, ".read": false } } }

Slide 54

Slide 54 text

Slide 55

Slide 55 text

{ "rules": { ".write": false, "lives": { "$live_id": { ".read": "root.child('alive_lives').child($live_id).val() === true", "messages": { "$message_id": { ".validate": "newData.hasChildren(['user', 'image', 'text'])" } }, "notifications": { "buy_item": { ".validate": "newData.hasChildren(['text'])" } } } }, "alive_lives": { ".write": false, ".read": false } } }

Slide 56

Slide 56 text

Slide 57

Slide 57 text

"root.child('alive_lives').child($live_id).val() === true”, Refer data on database

Slide 58

Slide 58 text

"root.child('alive_lives').child($live_id).val() === true”, Permission condition

Slide 59

Slide 59 text

{ "lives": { "1": { ... }, "2": { ... } }, "alive_lives": { "1": true, "2": false } } alive_lives[ʻ1ʼ] is true So it can be read

Slide 60

Slide 60 text

{ "lives": { "1": { ... }, "2": { ... } }, "alive_lives": { "1": true, "2": false } } alive_lives[ʻ1ʼ] is false So it canʼt be read

Slide 61

Slide 61 text

Easier to control permission • Updating rule every time will be overhead • So adding data to manage permission is good technique

Slide 62

Slide 62 text

How to run on production

Slide 63

Slide 63 text

Scalability Monitoring Troubleshooting What we needed

Slide 64

Slide 64 text

Scalability

Slide 65

Slide 65 text

Scalability • Mercari is high traﬃc service • Mercari channel also should think about for high traﬃc

Slide 66

Slide 66 text

KPI of Mercari channel • Over 1500 streamers/day • Over 30K viewers/day • 5-6M yen sales per 1 stream ※not strict numbers

Slide 67

Slide 67 text

Can Realtime Database be scale automatically?

Slide 68

Slide 68 text

No, we need to scale by ourselves

Slide 69

Slide 69 text

Again, limitation of database • There is limitation for each database • 1000 query/sec • 1M connection

Slide 70

Slide 70 text

Vertical sharding

Slide 71

Slide 71 text

subscribe Some data

Slide 72

Slide 72 text

subscribe Some data

Slide 73

Slide 73 text

For scalability • We use over 15+ database instances in production • Client switch DB by data from Mercari API • For now, Cloud Firestore can be good choice for this problem

Slide 74

Slide 74 text

Monitoring

Slide 75

Slide 75 text

Monitoring • In house monitoring tool • Stackdriver

Slide 76

Slide 76 text

In house • Application logging on Kibana • Queue status on Mackarel

Slide 77

Slide 77 text

Kibana

Slide 78

Slide 78 text

Mackerel

Slide 79

Slide 79 text

Stackdriver • Monitoring tool provided by Google • Some metrics of Realtime Database • Aggregating all instances data to 1 dashboard

Slide 80

Slide 80 text

Metrics • Connection count • Database load • Network usage • etc… https://cloud.google.com/monitoring/api/metrics_gcp#gcp-ﬁrebasedatabase

Slide 81

Slide 81 text

For our app • Connection count • Database load • REST API hit

Slide 82

Slide 82 text

Console

Slide 83

Slide 83 text

Troubleshooting

Slide 84

Slide 84 text

Only one truth

Slide 85

Slide 85 text

All services can be in trouble

Slide 86

Slide 86 text

What we need • Expect Cloud services as failing sometimes • Think about what we can for that

Slide 87

Slide 87 text

Notice Operation ・Notice to some problem ・Investigation ・Fix bug ・Make service maintenance Problem ・Some incident of Cloud service ・System alert

Slide 88

Slide 88 text

Notice Operation ・Notice to some problem ・Investigation ・Fix bug ・Make service maintenance Problem ・Some incident of Cloud service ・System alert

Slide 89

Slide 89 text

Problem from oﬃcial • There is oﬃcial status dashboard • RSS/JSON feed supported

Slide 90

Slide 90 text

Problem from system • Customized Stackdriver • Mackerel

Slide 91

Slide 91 text

Notice Operation ・Notice to some problem ・Investigation ・Fix bug ・Make service maintenance Problem ・Some incident of Cloud service ・System alert

Slide 92

Slide 92 text

Alert channel • All information will be passed to Slack channel • Channel includes all team members

Slide 93

Slide 93 text

Notiﬁcation from system • Stackdriver • Mackerel

Slide 94

Slide 94 text

Notiﬁcation from oﬃcial feed

Slide 95

Slide 95 text

When recovered

Slide 96

Slide 96 text

Small technique • Oﬃcial alert including both incident and recovery informations • There is a staﬀ who can't read English • So we customized feed bot

Slide 97

Slide 97 text

Customized message 4XJUDIFNPKJ PS

Slide 98

Slide 98 text

Customized message POMZNFOUJPOGPS JODJEFOU

Slide 99

Slide 99 text

Customized message "VUPUSBOTMBUJPO

Slide 100

Slide 100 text

Notice Operation ・Notice to some problem ・Investigation ・Fix bug ・Make service maintenance Problem ・Some incident of Cloud service ・System alert

Slide 101

Slide 101 text

Maintenance mode • System ﬂag to manage maintenance mode • Team member can edit • If the eﬀect of incident is too large for customers, we'll edit it

Slide 102

Slide 102 text

Appendix

Slide 103

Slide 103 text

Cloud Firestore vs Realtime Database • You SHOULD check this article • https://ﬁrebase.googleblog.com/2017/10/ cloud-ﬁrestore-for-rtdb-developers.html

Slide 104

Slide 104 text

For more detail https://speakerdeck.com/sota1235/realtime-messaging-with-ﬁrebase-number-phpcon2017

Slide 105

Slide 105 text

Thank you