No, I disagree. Have a look at the Dynamo paper — it only supports key-value lookups as opposed to local and global seondary indexing in DynamoDB, which supports real queries and filters, a schema, etc. Dynamo is also geared towards low latency (R=1, first answer wins) and write throughput (W=1, hinted hand-off) by the choice of the developer. DynamoDB claims low latency (by virtue of SSDs) but only for the inconsistent read mode. Furthermore it requires statically proviosed capacity units, which I deem the exact opposite of easy elasticity and scalability. The capacity requires continuous (usually manual) adjustment wheras in Dynamo servers can easily be added on the fly.