Amazon DynamoDB

Overview

DynamoDB is a database that appears quite a lot at the exam from an architectural point of view. It is fully managed, highly available, with replication of data across multiple AZs. It is cloud native (AWS proprietary) and a NoSQL database, so it is not a relational database like RDS and Aurora, although it still has transaction support.

Related: 02-AWS-Lambda (Lambda commonly uses DynamoDB as the database for an API), 10-DynamoDB-Advanced-Features.

Key Characteristics

You do not provision a database. The database is already there (the service is named DynamoDB). You create a table and say how much capacity you want in it.

Tables, Items and Attributes

• DynamoDB is made of tables. You never create a "database" like in Aurora or RDS.
• Each table has a primary key, decided at creation time.
• A table can have an infinite number of items (rows).
• Each item has attributes (like columns). Attributes can be added over time and can be null.
• Maximum item size is 400 KB. DynamoDB is not there to store very large objects.
• Supported data types: Scalar types (String, Number, Binary, Boolean, Null), plus List, Map and Set.
• The primary key is made of a partition key, and optionally a sort key.

Capacity Modes

You control how the table capacity is managed with one of two modes.

Hands-On Walkthrough

Create a table

1. Open DynamoDB and click Create table. There is no such thing as creating a database, you create a table right away, which makes it a serverless database.
2. Table name DemoTable. Choose a partition key (user_id) and optionally a sort key (none here). The combination of the two makes the primary key.
3. Choose Customize settings to see all options.

Read/write capacity settings

• On-demand: pay for the actual reads and writes. Great for unpredictable workloads, and AWS now recommends it as the default for most workloads.

• Provisioned: manage and optimize cost by allocating read and write capacity in advance. Read and write each have auto scaling on or off.
  • Auto scaling off: set fixed capacity units (e.g. 10 read and 5 write).
  • Auto scaling on: set a minimum (e.g. 1), a maximum (e.g. 100) and a target utilization (e.g. 70%). Auto scaling manages capacity to match about 70% utilization, for both read and write capacity.
• The demo sets auto scaling off with 1 RCU and 1 WCU.
• Indexes are covered in a later lecture, so they are skipped here.
• A capacity calculator exists to find the right read/write capacity. It is more complex and more relevant for the developer exam.
• With 1 RCU and 1 WCU the console shows an estimated cost of 71 cents per month, which gives pricing information.
• Enable encryption at rest, then click Create table.

Insert items

1. Open the table and click View items, then Create item.
2. Item 1: user_id = stephane_123, add attributes name (String) = Stephane Maarek, favorite_movie = Memento, favorite_number (Number) = 42.
3. Item 2: user_id = alice_456, name = Alice Doe, favorite_movie = Pocahontas, plus an extra attribute age = 23.
4. Alice has a different set of attributes than Stephane. Not all items need all attributes, every item can have different attributes. This shows the power of a NoSQL table versus SQL.

Questions & Answers