Do you know DynamoDB?

Amazon DynamoDB is a fully managed non-relational database service that provides fast and predictable performance with seamless scalability.

Use cases

• Mobile
• Web
• Gaming
• Advertisement
• IoT

Avoiding throttling

• Access is evenly spread over key-space
• Cache popular items
• Read/write sharding
• Requests arrive evenly spaced in time

Other features

• Auto scaling
• Metrics, monitoring, alarms
• Triggers & streams (AWS Lambda)
• Time to live attribute
• Atomic updates
• DAX: transparent, in-memory cache

Throughput & costs

• Read Capacity Unit (RCU) = 4 KB
• Write Capacity Unit (WCU) = 1 KB
• Billed by the hour

* Region South America, monthly price before taxes.

Source: http://calculator.s3.amazonaws.com/index.html

Concepts: Table

* Max item size: 400 KB

Concepts: Indexes

Table

Local Secondary Index

Global Secondary Index

Concepts: Indexes

• Sparse: ignores items missing key attributes
• Supported in Query & Scan operations
• Local
• Synchronous update
• Global
• Eventually consistent, always

Concepts: Data Types

• Simple
• string (except empty string)
• number (stored as text)
• binary
• Other
• boolean
• set (of string, number, binary)
• map
• list (of any)
• null

DynamoDB Math

Data Operations

Get Item

{
    TableName: "Users",
    Key: {
        id: "user-123"
    },
    ConsistentRead: false,
}
                        

Data Operations

Scan

{
    TableName: "Users",
    IndexName: 'ByCountry',
    FilterExpression: "country = 'Brazil'",
    Limit: 10,
}
                        

Data Operations

Query

{
    TableName: "Messages",
    KeyConditionExpression: "userId = 'user-123' and date > '2017-10-15'",
}
                        

Data Operations

Put Item

{
    TableName: "Messages",
    Item: {
        id: "msg-1",
        userId: "user-123",
        date: "2017-11-07",
        subject: "TDC is about to start...",
        body: "...",
    },
}
                        

Data Operations

Update Item

{
    TableName: "Game",
    Key: {
        id: 'game-123',
    },
    UpdateExpression: "
        set topLeft = 'O',
            version = version + 1,
            moves = list_append(moves, ['Alice']),
            metadata = if_not_exists(metadata, {})
        remove pending
        add 'pending' to gameState
    ",
    ConditionExpression: "version = 1",
}
                        

Data Operations

Delete Item

{
    TableName: "Messages",
    Key: {
        id: 'msg-123',
    },
}
                        

Update Expressions (SET)

• Functions: if_not_exists, and list_append

Update Conditionals

• Functions: ATTRIBUTE_EXIST, CONTAINS, and BEGINS_WITH
• Comparison: =, <>, <, >, <=, >=, BETWEEN, and IN
• Logical: NOT, AND, and OR

Modeling with DynamoDB

• How often data is read?
• How big are items?
• How data is retrieved?
• What attributes change on items?
• Any popular items?

1:N relations - I

Conference Track - talks as lists

Writes are expensive and non-atomic.

1:N relations - II

Conference Track - talks as maps

Writes are atomic. Writes are still expensive.

1:N relations - III

Conference Track - talks as separate table

Writes are cheap.

1:N relations - IV

Conference Track - composite sort keys

Reads are flexible.

Large Items - Overflow table

Good for read-only views.

Time Series

Events_2017_11

Events_2017_10

Events_2017_09

Time Series

Events

EventsByBucketIndex (GSI)

Scatter / Gather

Are you missing...

• Transactions?
• Joins?
• Ad hoc queries?
• Geospatial data?
• Backups?

Wait...

Did you say NO BACKUPS?

Next steps

1. Download DynamoDB Local at http://docs.aws.amazon.com/amazondynamodb/latest/developerguide/DynamoDBLocal.html
2. Run and navigate to http://localhost:8000/shell/
3. Type tutorial.start() and press ENTER
4. Follow steps

References

• https://aws.amazon.com/dynamodb/