What are the pros and cons of DynamoDB with respect to Google Cloud Datastore

Question

My understanding is DynamoDB behave like a giant table which you must specify a hash key and range key.

The core concept of Google Cloud Datastore is entity based (like Cassandra) and is more flexible, i.e. can use more than 1 index.

But are there any more in-depth comparison?

Answer 1

AWS DynamoDB is a pretty simple flat key-value store. It has support for conditional writes and sets which allow for some cool features. You specify the amount of horsepower you want (which you can only adjust a few times a day) and AWS splits up your dataset uniformly across enough database nodes to meet your demands. You have to make sure your key values are sufficiently random as to guarantee balanced access across your dataset. AWS almost guarantees single-digit latencies. Transactions are not supported. You specify the consistency of operations.

Google Cloud Datastore is a more sophisticated key-valueish store with built-in transaction support and entity hierarchy. You don't have to worry about the capacity of the system, it automatically scales to your data size and access patterns. You have less control of some things so you have to pay attention. You cannot specify for a read to be consistent, but you can force consistency by structuring your entities in a certain way.

One downside of Google Cloud products I have experienced is that documentation and language support is not very uniform. Sometimes you have to read documentation of another language to understand the system fully and many features are not supported in certain languages.

There are a lot of other differences. Look at the API reference of your favorite language on both documentation pages and you'll get a decent feel of the specific features of each.