Dynamo DB

NoSQL vs Relational

NoSQL:

• No Relations
• High Flexibiliy
• Data Repetition
• No integrity checks
• Easy Scalability

Relations:

• Relations
• Limited Flexibiliy (Strong Schemas)
• No Data Reletition
• Integrity Checks
• Harder Scalability

Need to manage the servers and clusters for relations.

How it works:

• NoSql Database
• No Relations

Data Format:

• Key/Value Pair
• No schema

Key, Attributes, Indexes:

Always required: Partition Key

• Has to be unique and be present on each item

Eg:

Partition Key: UserId

Attribute: FirstName

Why is it name Partition Key?

How Dynamo store the data, uses fleet of SSD, does this by partioning the state drive

Simple example: Store them from A B C ...to Z

If your userId starts with A1, A2, A3, it will get stored in A partition

Sort Key

Partition and a Sort Key as a Primary Key

Both don't have to be unique, but but when combined they are eg: UserId + timestamp

Query Data

When you need to query other attribute eg: FirstName

Global Secondary Index

For that you will need to set up a 'Global Secondary Index' (5 per table), this forces AWS to manage this attribute in an optinal way, when querying makes it more efficient.

Local Secondary Index

Provision Capacity

How often do you read/write this data, but also the size of item you are reading/writing

What about multiple databases?

You can only create table per regions