The Dynamo database is for non-relational uses that do not involve complex queries, joins or in-depth transactions. It is meant to support high numbers of users and interactions that might need to ramp to massive amounts of CPU and storage very quickly in a pay-as-needed model.
As a use case, Amazon offers the example of Internet applications such as online gaming, where a successful launch might quickly scale from hundreds to millions of users and create terabytes of data in the process.
Amazon CTO Werner Vogels says his company has used the database internally to run parts of its Web-based retail business. Vogels discussed the history and value proposition of the service in his blog; Amazon also offers an overview and developer guide at its website.
DynamoDB (Promotional) Overview
DynamoDB builds on the AWS SimpleDB service, but with what the company calls unlimited storage and fast predictable performance that runs completely on replicated solid-state disks. Like Amazon's SimpleDB service, Dynamo DB uses application programming interfaces to write and read data, and a management console to monitor status and performance.
As a fully-managed service, Dynamo DB covers multiple chores of hardware provisioning, partitioning, configuration and administration for any amount of capacity made available as needed. These functions are business-critical for Internet-based companies that will surely vet the service before putting it to production use.
And since it is a non-traditional database, DynamoDB will face a limited audience that the company hopes will grow as the service is approached and tested. The company is offering a free tier of 40 million read/write requests and 100 MB of storage per month, additional capacity at $0.01/hr and solid-state storage for $1 per GB per month.