Slide 1
Slide 1 text
Firestore Database Design
Session 2
#DevFest
Norikazu Muramoto @1amageek
Firebase Japan User Group Organizer
Slide 2
Slide 2 text
Norikazu Muramoto
@1amageek
Cookpad
Slide 3
Slide 3 text
System Device Technology Ltd,
2017.5
2015.10
2009.4
2018.6
Timers Inc
Cookpad Inc
Freelance
2018.9 Stamp Inc
LSI design engineer
iOS developer
Business development, Development Manager
CEO
Bio
Slide 4
Slide 4 text
Firebase Japan User Group
Slide 5
Slide 5 text
What is Cloud Firestore?
Slide 6
Slide 6 text
Cloud Firestore
Slide 7
Slide 7 text
Cloud Firestore
• Flexibility
• Expressive querying
• Realtime updates
• Offline support
• Designed to scale
SubCollection is an important feature of Cloud Firestore.
Slide 8
Slide 8 text
What is SubCollection?
Slide 9
Slide 9 text
Cloud Firestore Data Model
• Collection
• Document
• Data
Slide 10
Slide 10 text
Cloud Firestore SubCollection
A Document can have a SubCollection.
Slide 11
Slide 11 text
How do we design
products using Firestore?
Slide 12
Slide 12 text
Relationship Design
Important!
Slide 13
Slide 13 text
How to make a relationship
in Cloud Firestore?
• Key
• Reference
• Query
• SubCollection
• Same ID
We need to understand Client Side Join to understand the
NoSQL relationship.
There are several ways to create a relationship
in Cloud Firestore.
Slide 14
Slide 14 text
What is Client Side Join?
Slide 15
Slide 15 text
Server Side Join
Server
Data
Data
Data
JSON
In RDB, join on the server side.
Slide 16
Slide 16 text
Client Side Join
Cloud Firestore
Data
Client Side Join loads necessary data each time.
Snapshot
Data
Slide 17
Slide 17 text
Types of Relationship
Slide 18
Slide 18 text
Relationship by Key
ID: xxx
child: yyy
ID: yyy
parent: xxx
Key Relationship is used for cross-referencing.
However, since Document only knows the ID of the
reference destination, the developer must manage the path.
/parent/xxx /child/yyy
Slide 19
Slide 19 text
Relationship by Reference
ID: xxx
child: yyy
ID: yyy
parent: xxx
Reference Relationship is used for cross-referencing.
It holds the reference path.
/parent/xxx /child/yyy
/child/yyy /parent/xxx
Slide 20
Slide 20 text
Relationship by
Query, SubCollection
• Reference
The snapshot acquired by the query does not
contain data.
• Value
It contains data in the snapshot acquired by the
query.
There are two kinds of relationship by Query, SubCollection.
Slide 21
Slide 21 text
Relationship by Query
ID: 0
parent: yyy
ID: 1
parent: xxx
ID: 2
parent: xxx
ID: 3
parent: yyy
/child/ Here, the relationship between
parent and child is associated by
ID.
firestore
.collection(“child”)
.where(equalTo: “xxx”)
Slide 22
Slide 22 text
Relationship by Query
ID: yyy
data: { }
/junction/
ID: 0
parent: yyy
child: xxx
ID: xxx
data: { }
/parent/
/child/
This expresses relationships through junction collections.
Slide 23
Slide 23 text
ID: xxx
data: { }
ID: yyy
data: { }
/parent/ /parent/child/
Relationship by SubCollection
The SubCollection itself represents a relationship.
Slide 24
Slide 24 text
ID: xxx
data: { }
ID: yyy
data: { }
ID: yyy
/parent/
/child/
/parent/child/
Reference
Relationship by SubCollection
Create a relationship by having the reference ID in
SubCollection.
Slide 25
Slide 25 text
ID: xxx
data: { }
ID: xxx
data: { }
/parent/:same_id
/child/:same_id
Relationship by Same ID
By using the same ID for another pass, we express the
relationship between the two documents.
Slide 26
Slide 26 text
How to use it?
Slide 27
Slide 27 text
Security Rules
To understand how to use, we need to know
about security rules.
Slide 28
Slide 28 text
Security Rules
• It is not possible to limit individual fields.
It is necessary to separate documents between
secure information and non-secure information.
• When reading limits are applied, the query
cannot be used.
It is necessary to use SubCollection to create a
secure collection.
I will explain only the necessary points here. It takes time to
explain security rules in detail.
Slide 29
Slide 29 text
Key
Key is good for representing relations with a single object.
Slide 30
Slide 30 text
Reference
Reference is suitable for a relationship having the same ID and
having a different path.
Slide 31
Slide 31 text
Same ID
It has a different structure with the same ID.
Same ID is suitable for separating public information from private
information.
Slide 32
Slide 32 text
Field Query
It is the same as `where` Query of RDB.
Field Query is suitable for expressing relationships of published
information.
Slide 33
Slide 33 text
SubCollection
SubCollection is suitable for expressing relationships of
information that we want to maintain security.
Slide 34
Slide 34 text
Junction Collection
Junction Collection is suitable for the relationship when there are
two information roles.
Slide 35
Slide 35 text
Reference Collection
The Reference Collection is suitable for the relationship between
the followee and follower.
Slide 36
Slide 36 text
NoSQL has some techniques
that you should not forget.
Duplicate
Slide 37
Slide 37 text
What is Duplicate?
Slide 38
Slide 38 text
Duplicated Data
ID: yyy
data: { }
/parent/item/
ID: 0
parent: yyy
child: xxx
ID: xxx
data: { }
/parent/
/child/
Writes exactly the same data to the two references.Writes exactly
the same data to the two references.
By doing this, JOIN becomes unnecessary, and data can be
accessed at high speed.
/child/item/
ID: 0
parent: yyy
child: xxx
Slide 39
Slide 39 text
Duplicated Collection
Duplicated Collection is suitable for relationships that are read
frequently, but are less frequent to update.
Slide 40
Slide 40 text
1 : 1
1 : N
N : N
Key Reference
Field Query SubCollection ( Nested Collection )
Junction Collection Reference Collection
Same ID
Cloud Firestore Relationships
Duplicated Collection
Slide 41
Slide 41 text
Thank you.