This looks like a great addition to Amazon S3 and EC2 - they now have a way of creating simple tables with up to 256 attribute/value pairs. Support for GET, PUT, DELETE, and QUERY.

There are a number of things missing over a real relational database (or even an ISAM). S3 is a simple "blob" store - where each object get be atomically read and written. SimpleDB gives some simple query capability (for smaller data items - 1K max per attribute).

I also see no way to get ORDERBY functionality and, of course, no aggregate values (SUM, MAX, etc). Sometimes this can be overcome in the application layer, but it will make your application more complex than using a traditional database.

Most importantly, there are no JOIN's in this system. This will require the application to make multiple QUERYs to get the data needed to generate a view.

I found it very interesting that this document states that Amazon promotes developers running traditional databases in EC2. My understanding is that Amazon can arbitrarily wipe the state of any machine on EC2 - which would cause you to have to restore the database from some an S3 or external backup.

Maybe someone who is using EC2 for running a database server can chime in on how well that works for them.

Quoted: Amazon SimpleDB - a web service for running queries on structured data in real time. Provides the core functionality of a database - real-time lookup and simple querying of structured data - without the operational complexity