How does SimpleDB work?

The best way to wrap your head around the way SimpleDB works is to picture a spreadsheet that contains your structured data. For instance, a contact database that stores information on your customers can be represented in SimpleDB as follows:

As SimpleDB is a different database metaphor, new terms have been introduced. The use of this new terminology by Amazon stresses that the traditional assumptions may not be valid.

Domain

The entire customers table will be represented as the domain Customers. Domains group similar data for your application, and you can have up to 100 domains per AWS account. If required, you can increase this limit further by filling out a form on the SimpleDB website. The data stored in these domains is retrieved by making queries against the specific domain. There is no concept of joins as in the relational database world; therefore, queries run within a specific domain and not across domains.

Item

Each customer is represented by a unique Customer ID. Items are similar to rows in a database table. Each item identifies a single object and contains data for that individual item as a number of key-value attributes. Each item is identified by a unique key or identifier, or in traditional terminology, the primary key. SimpleDB does not support the concept of auto-incrementing keys, and most people use a generated key such as the unix timestamp combined with the user identifier or something similar as the unique identifier for an item. You can have up to one billion items in each domain.

Attributes

Each customer item will have distinguishing characteristics that are represented by an attribute. A customer will have a name, a phone number, an address, and other such attributes, which are similar to the columns in a table in a database. SimpleDB even enables you to have different attributes for each item in a domain. This kind of schema independence lets you mix and match items within a domain to satisfy the needs of your application easily, while at the same time enables you to take advantage of the benefits of the automatic indexing provided by SimpleDB. If your company suddenly decides to start marketing using Twitter, you can simply add a new attribute to your customer domain for the customers who have a Twitter ID! In traditional database terminology, there is no need to add a new column to the table.

Values

Each customer attribute will be associated with a value, which is the same as a cell in a spreadsheet or the value of a column in a database. A relational database or a spreadsheet supports only a single value for each cell or column, while SimpleDB allows you to have multiple values for a single attribute. This lets you do things such as store multiple e-mail addresses for a customer while taking advantage of automatic indexing, without the need for you to manually create new and separate columns for each e-mail address, and then index each new column. In a relational DB, a separate table with a join would be used to store the multiple values. Unlike a delimited list in a character field, the multiple values are indexed, enabling quick searching.

It is a simple way of modeling your data, but at the same time, it is different from a relational database model that is familiar to most users. The following table compares SimpleDB components with a spreadsheet and a relational database:

Relational Database	Spreadsheet	SimpleDB
Table	Worksheet	Domain
Row	Row	Item
Column	Cell	Attribute
Value	Value	Value(s)