Advanced design patterns for DynamoDB including strategies for Global Secondary Index sharding and index overloading, scaleable graph processing with materialized queries, relational modeling with composite keys, executing transactional workflows on DynamoDB, and more.
Write sharding spread writes for a hot partition key across multiple logical partitions by appending a random value to the partition key. Scatter-gather techniques can be used to aggregate values from multiple partitions, as can DynamoDB Streams.
A hi-res version of this images can be found here.
A discussion on write sharding begins at 12:07