Eventual consistency
Simple DB can be thought of as a Write-Seldom-Read-Many model. Updates are done to a central database, but reads can be done from many read-only database slave servers.
SimpleDB keeps multiple copies of each domain. Whenever data is written or updated within a domain, first a success status code is returned to your application, and then all the different copies of the data are updated. The propagation of these changes to all of the nodes at all the storage locations might take some time, but eventually the data will become consistent across all the nodes.
SimpleDB provides this assurance only of eventual consistency for your data. This means that the data you retrieve from SimpleDB at any particular time may be slightly out of date. The main reason for this is that SimpleDB service is implemented as a distributed system, and all of the information is stored across multiple physical servers and potentially across multiple data centers in a completely redundant manner....