What approach will result in the simplest, decoupled, and reliable method to get near-real time updates from the Accounts database?
Use Amazon Glue to perform frequent ETL updates from the Accounts database to the Payments database.
Use Amazon ElastiCache in Payments, with the cache updated by triggers in the Accounts database.
Use Amazon Kinesis Data Firehose to deliver all changes from the Accounts database to the Payments database.
Use Amazon DynamoDB Streams to deliver all changes from the Accounts database to the Payments database.
Explanations:
Amazon Glue is typically used for batch ETL (Extract, Transform, Load) jobs, not for real-time or near-real-time updates. It would introduce latency and increase complexity, which goes against the goal of decoupling and providing real-time updates.
Amazon ElastiCache is a caching solution and is not suitable for real-time data synchronization between databases. It would add complexity with the need for cache invalidation and does not directly address the need for real-time updates from the Accounts database.
Amazon Kinesis Data Firehose is more suited for streaming data delivery to destinations like S3, Redshift, or Elasticsearch. It is not designed to directly update DynamoDB, and it may not provide the level of control or granularity needed for updating a DynamoDB table.
Amazon DynamoDB Streams provides a reliable, real-time mechanism for capturing changes to DynamoDB tables. It can easily deliver updates from the Accounts database to the Payments database with minimal delay and without introducing complex infrastructure. This option is the simplest and most decoupled approach for near-real-time updates.
As I understand it, the answer is:
Use Amazon DynamoDB Streams to deliver all changes from the Accounts database to the Payments database.