[D3T2S03] Data&AI Roadshow 2024 - Amazon DocumentDB 실습

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
AWS DATA & AI ROADSHOW 2024
Jun Kim
Database Expert Principal SA
AWS DocumentDB
Hands-on Session
Hyojeong Han
TAM

2
Agenda
• What is Amazon DocumentDB?
• DocumentDB Overall Architecture
• DocumentDB Replication
• 1st hands-on
• DocumentDB Modeling
• 2nd hands-on
• Q&A

3
Amazon DocumentDB (with MongoDB compatibility)
Fully managed and scalable
document database service that
supports MongoDB workloads
Scalable
Fully managed
MongoDB API
compatible

Backups enabled by default
Durable by default
Built-in high availability
Security best practices by default
Automatic patching
Monitoring and alerting
Fully managed

Scalable
Scale compute in minutes
Storage and IO autoscaling
Storage scales to 128TiB
Scale out to 15 replicas for millions of reads

6
Applications, drivers, and tools can be used with
Amazon DocumentDB with little or no change
Supports hundreds of APIs, operators, and stages
Continually working backward from customers
to deliver the capabilities they need
MongoDB API
compatible

7
When shoud you use a document database?
Amazon
DocumentDB makes
it easy to
store, query, and
index JSON data
JSON data
Operational and
analytics workloads
Ad hoc query
capabilities
Flexible indexing
Flexible schema for
fast iteration

Architecture

9
DocumentDB Architecture
Instance
(replica)
Reads
Instance
(primary)
Reads
Writes
Instance
(replica)
Reads
Distributed storage volume
db.foo.find({}) {"x":1}
AZ 1 AZ 2 AZ 3

10
Separation of
storage / compute
How would you
build a cloud-native
database
architecture?
Instance
(replica)
Reads
Instance
(primary)
Reads
Writes
Instance
(replica)
Reads
AZ 1 AZ 2 AZ 3
Compute
Storage

11
Separation of
storage / compute
API
Query processor
Caching
Logging
Storage
Log writes
Decouple compute and storage
Compute layer
Storage layer
Separation of
storage and
compute

12
Separation of
storage / compute
AZ 1 AZ 2 AZ 3
Compute
Storage
Instance
(primary)
Reads
Writes
r6g.large

13
Separation of
storage / compute
AZ 1 AZ 2 AZ 3
Compute
Storage
Instance
(primary)
Reads
Writes
r6g.large
Instance
(replica)
Reads
r6g.large

Replication

15
DocumentDB Replication
Replication
Instance
(replica)
Reads
Instance
(primary)
Reads
Writes
Instance
(replica)
Reads
AZ 1 AZ 2 AZ 3
Compute
Storage

16
Replication
db.foo.insert({’x’:1})
Instance
(replica)
Reads
Instance
(primary)
Reads
Writes
Instance
(replica)
Reads
AZ 1 AZ 2 AZ 3
Compute
Storage

17
Replication
Instance
(replica)
Reads
Instance
(primary)
Reads
Writes
Instance
(replica)
Reads
AZ 1 AZ 2 AZ 3
Compute
Storage
ACK

18
Replication
db.foo.insert({’x’:1}) ACK
Instance
(replica)
Reads
Instance
(primary)
Reads
Writes
Instance
(replica)
Reads
AZ 1 AZ 2 AZ 3
Compute
Storage
Eventual
consistency
Eventual
consistency

19
Replication
ACK
db.foo.insert({’x’:1}) ACK
Instance
(replica)
Reads
Instance
(primary)
Reads
Writes
Instance
(replica)
Reads
AZ 1 AZ 2 AZ 3
Compute
Storage
Eventual
consistency
Eventual
consistency

20
Replication
Instance
(replica)
Reads
Instance
(primary)
Reads
Writes
Instance
(replica)
Reads
AZ 1 AZ 2 AZ 3
Compute
Storage
Eventual
consistency
Eventual
consistency
db.foo.find({}) {‘x’:1}

1st Hands-on
• Create DocumentDB Cluster
• CRUD
• Scale Cluster

Modeling

23
Describe the application requirements
1. Workload Define
• Estimate inventory of 100,000 products, in the first year (product data to be kept forever)
• Estimate a number of 10,000 customers, in the first year (customer account data to be kept forever)
• Order and reviews data needs to be kept for 5 years.
• Estimated read/write ratio is 80% / 20%
2. Read & Write Queries
• Insert new products
• Update existing products
• Insert customers
• Insert reviews added by customers
• Insert orders
• Update orders
• Read products
• Read orders
• Read reviews
• Read customer data
• Run reports
Example

24
CRUD Frequency Type
New Products
Added
300/day Write
Product views 6500/sec Read
New customer
added
30/day Write
Customer logs in to
website
30 user logins/sec Read
New order added 10/hour Write
New review added 5/hour Write
Run reports 1/day Read
op Desc Type
Max
latency
Avg Freq
/sec
Max Freq
/sec
w1 New Product added or
updated
I/U < 500ms 5 10
w2 Customer creates
account
I < 100ms 1 3
w3 New review added for
product
I < 200ms 3 12
w4
Customer creates order I 10ms 12 30
w5 Customer adds
products to order
U 20ms 16 48
R1
Customer logs into app R 5ms 32 64
R2 Customer views a
specific product
R 1m 250 6500
R3 Customer views their
orders
R
20ms 20 80
R4 Analytics report
executed
R < 300sec <1 2
R5
Customer views review R 5ms 12 36

25
ü List the actual requirements of the application.
This is where you define what the application is supposed to do and what data will be stored in the
database
ü Estimate the data size
ü Quantify the operations, such as the total queries ran against the database, and how many reads
versus writes are expected
ü Qualify those operations, think about the most important queries and latency requirements of those
queries
ü Identify consistency requirements and tolerance to stale data
- Key Takeaways : Find out how the data will be used for workload.

26
Identify Relationship
- Think about the relationship between collection ( Reference or Embed )
Relationship :
• One-to-one
• One-to-many
• Many-to-many

27
- Think about the relationship between collection ( Reference or Embed )
Referenced vs Embed

28
- Referencing
Procedure
{
"_id" : 333,
"date" : "2003-02-09T05:00:00"),
"hospital" : “County Hills”,
"patient" : “John Doe”,
"physician" : “Stephen Smith”,
"type" : ”Chest X-ray",
”result" : 134
}
Results
{
“_id” : 134
"type" : "txt",
"size" : NumberInt(12),
"content" : {
value1: 343,
value2: “abc”,
…
}
}

29
- Embedding
Procedure
{
"_id" : 333,
"date" : "2003-02-09T05:00:00"),
"hospital" : “County Hills”,
"patient" : “John Doe”,
"physician" : “Stephen Smith”,
"type" : ”Chest X-ray",
”result" : {
"type" : "txt",
"size" : NumberInt(12),
"content" : {
value1: 343,
value2: “abc”,
…
}
}
}

30
- Referencing vs Embedding
Pros vs Cons

31
Apply Design Patterns
- Attribute pattern
Before Apply Pattern After Apply Pattern
{
“_id”: <objectId>,
“productid”: <productid>,
“name”: <string>,
“description”: <string>,
“size”: <string>,
“weight”:<int>,
“color”:<string>,
“packaging”:<string>
}
{
“attributes”: [
{“key”: ”size”, “value”:<string>},
{“key”: ”weight”, “value”:<int>},
{“key”: ”color”, “value”:<string>},
{“key”: ”packaging”, “value”:<string>},
]
}
v Benefits
Leverage Indexing - attributes.key
Easy to expand qualifier
• { descriptor: "price", qualifier: "euros", value: Decimal(100.00) }
{ descriptor: "price", qualifier: “won”, value: Decimal(80000.00) }

32
- Bucket pattern
{
“attributes”: [
{“key”: ”size”, “value”:<string>},
{“key”: ”weight”, “value”:<int>},
{“key”: ”color”, “value”:<string>},
{“key”: ”packaging”, “value”:<string>},
]
}
v Benefits
Practical benefits of the document model ( 1:N )
Reduce indexing size
Increased speed in extracting relevant data

33
- Subset pattern
v Benefits
Maintain small working set ( increasing cache efficiency )
Read Performance Increase

34
One Takeaway
- you must remember
Query Together
Should be
Saved Together !

2nd Hands-on
• How to model the document

QnA = True
if QnA:
doQuestions()
else:
doSurvey()

Thank you!

[D3T2S03] Data&AI Roadshow 2024 - Amazon DocumentDB 실습

More Related Content

Similar to [D3T2S03] Data&AI Roadshow 2024 - Amazon DocumentDB 실습

Similar to [D3T2S03] Data&AI Roadshow 2024 - Amazon DocumentDB 실습 (20)

More from Amazon Web Services Korea

More from Amazon Web Services Korea (20)

Recently uploaded

Recently uploaded (20)

[D3T2S03] Data&AI Roadshow 2024 - Amazon DocumentDB 실습