I have a DynamoDB table that changes quite frequently (INSERTs, UPDATEs and DELETEs).
I want to reflect those changes into an RDS table.
I saw that some articles that suggest the following architecture:
DynamoDB -> DynamoDB Stream Processor lambda -> (Kinesis Firehose <-> Transform Lambda)
-> S3 (transformed and untrasformed buckets) -> Lambda -> Postgres RDS.
It made me wonder:
- Why do I need the Kinesis adapter in the middle?
- Can I get a good-enough solution (1,000~ changes per second, or if possible even 10,000 changes per second) with another architecture, perhaps simpler one?