Is there an alternative to logged batches on Amazon Keyspaces?

Question

Our team decided to move with Amazon Keyspaces from self hosted Cassandra. After configured with Amazon Keyspaces I found that logged batches are not supported. In our solution there are multiple cases where it requires to use logged batches.

Ex: We have few tables where we are propagating data, in order to have different primary keys for fast querying. Logged batches use here to maintain consistency in all tables.

Is there any alternative solution to this type of scenario in Amazon Keyspace ?

Answer 1

Unfortunately with the lack of support for LOGGED BATCH in Amazon Keyspaces, you would somehow need to implement the logic within your application to keep the tables synchronised.

You would need to retry any failed updates to a table and if after N attempts it is still unsuccessful, you need to work out whether you need (a) issue deletes to the other tables, or (b) continue retrying.

It is unfortunate that you weren't aware of this limitation in Amazon Keyspaces. I would recommend familiarising yourself with other limitations in the following documents:

• Functional differences with Apache Cassandra
• Supported Cassandra APIs, operations, functions, and data types